Repetitive Behaviours, Anxiety and Sensory Problems in Children with Autism and Correlates of Anxiety in their Parents

Transcription

1 Wales Autism Research Centre, Cardiff University, School of Psychology, Cardiff, Wales Repetitive Behaviours, Anxiety and Sensory Problems in Children with Autism and Correlates of Anxiety in their Parents Mirko Uljarević A Thesis Submitted to the School of Psychology For the Degree Doctor of Philosophy 2013

2

3 CONTENTS Acknowledgements List of Presentations Thesis Summary List of Tables List of Figures Chapter 1: General Introduction Repetitive Behaviours Summary of factor analytic studies that used ADI-R How do repetitive behaviours in autism change over time? Repetitive Behaviours in Typical Development Repetitive Behaviours in other neurodevelopmental and psychiatric conditions Distinctiveness of repetitive behaviours in autism when compared with other clinical groups Distinctiveness of RRBs as a core impairment or the relationship between repetitive behaviours and other core autism symptoms Summary Anxiety and Autism Prevalence of anxiety in autism Moderators of anxiety Relationship between core autism symptomatology and anxiety... 46

4 Theories of Anxiety in autism Main issues in the measurement of anxiety in population with autism Summary Sensory Problems in Autism Sensory Discrimination Disorder Sensory Modulation Problems in autism Prevalence of Sensory Modulation Disorders in Autism The relationship between sensory modulation problems, chronological age and developmental level Sensory Problems and Core Autism Symptoms Summary Repetitive behaviours, sensory problems and anxiety in autism Repetitive behaviours and sensory problems Repetitive Behaviours and Anxiety Sensory Problems and Anxiety Summary Chapter 2: General method Section Background Measures Selection of the questionnaires Description of questionnaires... 71

5 2. 3. Recruitment process Ethics Recruitment Procedure Description of the primary sample (South Wales Sample) Description of the secondary data sets Secondary Data Set Secondary Data Set Methodological Limitations of this thesis Chapter 3: Repetitive Behaviours in Autism and Typical Development Chapter Plan: Introduction Measurement of Repetitive Behaviours Study 1 Developmental change in RRBs in typical development Methods Results Study 2 Repetitive Behaviours in Autism Methods Results Discussion Developmental trajectory of RRBs in TD children Moderators of repetitive behaviours in autism

6 Limitations Summary Chapters 4 and 5: Anxiety and Sensory Problems in children and adolescents with autism Chapter 4: Anxiety in children and adolescents with autism Chapter Plan: Introduction Assessment of anxiety in Autism Evaluation of the anxiety instruments Objectives Methods Subjects Measures Results Psychometric Properties of Anxiety Instruments Prevalence of Anxiety Relationship between anxiety levels and other children s variables Discussion Chapter 5: Sensory Processing Problems in Autism Chapter Plan: Introduction Objectives

7 5. 2. Methods Subjects Measures Results Frequency of Sensory Processing problems across different SP categories Relationship between the sensory quadrants Moderators of Sensory Modulation Problems Discussion Summary of the findings on Anxiety and Sensory Processing in Autism Limitations of the findings on anxiety and sensory processing in autism Chapter 6: Repetitive Behaviours, Anxiety and Sensory Problems in Autism Chapter Plan: Introduction Problem in arousal modulation as a key factor for the repetitive behaviours Sensory problems and anxiety as triggers for repetitive behaviours Summary and aims Methods Subjects Measures Results Discussion

8 Summary and limitations Chapter 7: Anxiety in Parents of Children with Autism: Prevalence and Risk factors Prevalence of anxiety in parents of children with autism Characteristics of children with autism as a contributing factor to anxiety in their parents Characteristics of parents as a contributing factor to their own levels of anxiety _Toc Broader Autism Phenotype (BAP) as a risk factor for the development of anxiety in parents Coping in parents of children with autism as a protective/risk factor for the development of anxiety Characteristics of the environment as contributing factors to the development of anxiety Intolerance of Uncertainty Sensory Processing in Parents of Children with Autism Relationship between Intolerance of Uncertainty, Sensory Sensitivity and Coping Styles Chapter 8: Prevalence of Anxiety in Parents of children with autism and contribution of children s, parents and environmental characteristics Chapter Plan: Introduction Prevalence of anxiety in parents of children with autism Risk factors for anxiety in parents of children with autism

9 Summary of the chapter aims Methods Subjects Measures Results Frequency of anxiety in parents Relationship between anxiety and children s variables Relationship between parent s own characteristics and their levels of anxiety Relationship between environmental characteristics and anxiety in parents Discussion Limitations Summary Chapter 9: Intolerance of Uncertainty and Sensory Problems in Children with Autism and their Contribution to Anxiety Levels in Parents Chapter Plan: Introduction Intolerance of uncertainty and anxiety Sensory atypicalities and their relationship with anxiety Intolerance of uncertainty, sensory atypicalities and coping Notes on the measurement of Sensory Problems and Intolerance of uncertainty Summary of the chapter aims

10 9. 2. Methods Subjects Measures Results Frequency of sensory problems in parents of children with autism The relationship between Sensory problems in children and mothers Relationship between sensory atypicalities and anxiety The relationship between anxiety, intolerance of uncertainty and sensory problems in mothers of children with autism Predictors of escape-avoidance coping in mothers of children with autism Interrelationship between avoidance coping, anxiety and ease of excitation Discussion Limitations Summary Chapter 10: General Discussion Interrelationship between repetitive behaviours, sensory problems and anxiety Why is the interrelationship between repetitive behaviours, sensory problems and anxiety an important area of research? What were the aims and what has this programme of research achieved? Correlates of anxiety in parents Why is it important to understand anxiety problems in parents of children with

11 autism? What were the aims and what has this programme of research achieved? Limitations Implications and Future Directions Conclusions References Appendices Appendix 1 The Relationship between RRBs and other characteristics in individuals with autism _Toc Appendix 2 - Summary of the most widely used questionnaire and interview measures of repetitive behaviours Appendix 3 - Summary of the identified anxiety instruments Appendix 4 Background Questionnaire

12 Acknowledgements I would like to dedicate this thesis to my mother and father, Mirjana and Zoran, to my sister Miljana, and to Milena. Words fail me to express how great they were through all of stages of my PhD, I just hope they know how grateful I am for all their support and understanding and for just being themselves really. I am endlessly grateful to my supervisor Sue, without whose her help I wouldn t have been able to start my PhD journey and would definitely not be able to achieve even a fraction of what I have achieved for the past three years. She was always very supportive of my research curiosity and provided much needed guidance at every stage of my PhD and I can t imagine having a better mentor. I am also grateful to my second supervisor prof. Andrew Lawrence for his always helpful advice. I conducted my research project within the Wales Autism Research Centre and everyone was always really helpful and understanding and made my PhD much easier and more enjoyable. I d like to thank past and current members, they have all been equaly great: Linda, Anastasia, Lynda, Chris, Louise and Bev. I am especially grateful to Rachel for always being positive and supportive and to Sarah whose advice and encouragement throughout my PhD made it possible to get to this stage. I had shared office with some amazing people that have always been true friends, in particular with Dr Kate Ellis-Davies and Dr Scott Jones. I would like to thank all the families that took part in this research project. I am also grateful to the National Institute for Social Care and Health Research (NISCHR) and to the School of Psychology, Cardiff University for funding my PhD project.

13 List of Presentations Uljarević, M., Lidstone, J., Leekam, S., Freeston, M., Le Couteur, A., McConachie, H., Rodgers, J., & Prior, M. (2013). Correlates of anxiety in mothers of children with autism spectrum disorder. The European Academy of Childhood Disability annual meeting. Uljarević, M., Carrington S., Evans, D., Leekam, S. (2013). Anxiety, Temperament and Autistic Traits in the General Population. Talk given at the Anxiety premeeting, the International Meeting for Autism Research, San Sebastian, Spain. Uljarević, M., Lidstone, J., Leekam, S., Freeston, M., Le Couteur, A., McConachie, H., Rodgers, J., & Prior, M. (2012). Intolerance of Uncertainty and Coping Strategies As Correlates of Anxiety in Mothers of Children with ASD. Poster presented at the International Meeting for Autism Research (IMFAR), San Sebastian, Spain. Uljarević, M., Lidstone, J., Leekam, S., Freeston, M., Le Couteur, A., McConachie, H., Rodgers, J., & Prior, M. (2012). Sensory processing and repetitive motor behaviours in children with Autism Spectrum Disorders (ASD). Poster presented at the British Psychological Society Developmental Section Annual Conference, 5 th -7 th September, 2012 University of Strathclyde, Glasgow. Uljarević, M., Lidstone, J., Leekam, S. R., Kanaris, H., McKigney, A. M., Mullis, J., Paradice, R. (2012). Anxiety in Children with Autism Spectrum Disorders is Associated with Affective Symptoms in Their Mothers. Poster presented at the International Meeting for Autism Research (IMFAR), Toronto, Canada. Lidstone, J., Uljarević, M., Leekam, S.R., Kanaris, H., McKigney, A.-M., Mullis, J., & Paradice, R. (2012). How do the functions of restricted and repetitive behaviours vary with developmental level in children with autism spectrum disorders? Poster presented at the International Meeting for Autism Research, Canada, Toronto, Canada. Uljarević, M., Lidstone, J., Leekam, S. R. (2012). Sensory symptoms and anxiety: A study of children with ASD and their parents. Talk given to the Bristol Autism Research Group (BARG), School of Experimental Psychology, Bristol. Uljarević, M. (2012). Sensory problems, anxiety and repetitive behaviours in children with ASD. School of Psychology, Cardiff University, internal seminar. Uljarević, M. (2011). The relationship between Anxiety, Sensory Processing and Repetitive

14 Behaviours in children and adolescents with ASD. Postgraduate Student Conference, Gregynog. List of Publications Lidstone, J*., & Uljarević, M*a, Sullivan, J., Rodgers, J., McConachie, H., Freeston, M., Le Couteur, A., Prior, M., & Leekam S. (2014). Relations among restricted and repetitive behaviours, anxiety and sensory symptoms in individuals with autism spectrum disorders. Research in Autism Spectrum Disorders, 8, Uljarević, M., Prior, M., & Leekam, S. First evidence of sensory atypicality in parents of children with Autism Spectrum Disorder (ASD). Molecular Autism. Under Review. Uljarević, M., Carrington, S., Arnott, B., Meins, E., McConachie, H., Le Couteur, A., Leekam, S. R. Developmental change in restricted and repetitive behaviours in young children. Journal of Child Psychology and Psychiatry. In preparation. Uljarević, M., Prior, M., & Leekam, S. Intolerance of Uncertainty and Coping Strategies As Correlates of Anxiety in Mothers of Children with ASD. Autism Research. In Preparation. *Joint First Authors; a Corresponding Author

15 Thesis Summary Repetitive and restricted behaviours have been considered as a core symptom of autism since the first descriptions provided by Leo Kanner (Kanner, 1943) and Hans Asperger (Asperger, 1944), and this view has been supported through all the incarnations of international diagnostic systems. However, when compared with other core features: problems in communication and problems in reciprocal social interaction, repetitive behaviours have been relatively neglected in terms of research. In particular, little is known about the triggers for repetitive behaviours and what functions they might serve for individuals with autism. Early theoretical accounts of autism proposed that sensory problems and anxiety are the key triggers for repetitive behaviours. Research in the field of developmental psychology has also demonstrated that in very young children, repetitive behaviours serve the function of constraining sensory unpredictability and warding of anxiety and this relationship has also been found in individuals with Obsessive Compulsive Disorder, a condition defined by anxiety and ritualism. However, the three way relationship between repetitive behaviours, sensory problems and anxiety has not been explored in the autism literature before. The main aim of the thesis (Part 1) was to explore this relationship in children with autism. In addition to this main focus, the work of the thesis also included a study of the mothers of the same children (Part 2), investigating mothers own anxiety, sensory correlates of this anxiety and other factors. The main effects of autism are on the individual who is on the autism spectrum, however, impact is also felt by families. Parents of children with autism have been found to have higher prevalence of affective disorders when compared to both parents of typically developing children and parents of children with other disabilities. The second main aim of the thesis was to explore correlates of anxiety in mothers of children with autism. In Part 1, a series of studies are reported using questionnaire data to investigate repetitive behaviours (RRBs), anxiety and sensory problems in children with autism. Part 2, focused on anxiety in the mothers of these children and correlates of their anxiety. The first empirical chapter (Chapter 3) examined RRBs in children with and without autism and validated the Repetitive Behaviour Questionnaire-2 (RBQ-2; Leekam et al., 2007) for the first time with children with autism. Results indicated that the RBQ-2 provides a highly reliable measure of repetitive behaviours for children with autism aged from 2 to 17 years. Factor analysis revealed that a two-factor structure with the repetitive motor and sensory behaviour (RSM) factor and the insistence on sameness (IS) factor best represented

16 the data. Internal consistency was high for both the total RBQ-2 scale and for each RSM and IS subscale separately. This chapter also explored moderators of repetitive behaviours in children and adolescents with autism and found that while RSM behaviours were negatively associated with lower chronological age, and lower expressive language levels, IS behaviours were not. Further analysis was conducted using a secondary dataset from a longitudinal study of RRBs in typically developing children at age 15, 24 and 72 months old. Results showed that IS and RSM behaviours represent relatively independent classes of behaviours in developmental terms. The second empirical chapter (Chapter 4) examined anxiety in children with autism and provided validation for the school age and preschool version of the Spence Anxiety Scales. Using this scale it was found that as many as 49% of children with autism met the criteria for elevated overall anxiety and that separation anxiety and physical injury fears were the most prevalent anxiety subtypes. Good agreement on total anxiety scores between children s and parents reports was found. Analysis on the mediators of anxiety suggested that anxiety was not associated with chronological age, impairments in communication and social interaction, and expressive language levels. The third empirical chapter (Chapter 5) provided evidence that sensory problems in children and adolescents with autism are both multisensory and multimodal in nature. More precisely, only 2 children had problems in a single sensory modality while on the other hand, 40% of children had problems simultaneously across all five primary sensory modalities. Almost 91% of children and adolescents showed mixed types of sensory modulation problems, with 65.3% of children having problems in all four sensory quadrants (sensory hyper- and hypo-sensitivity, seeking, and avoidance). This chapter also explored the mutual relationship between the four sensory subscale (quadrants) of the Sensory Profile. Results indicated that children and adolescents with autism fluctuate between the states of hypo- and hyper-responsiveness; that avoidance behaviours are compensatory strategies related to being over-stimulated, and that seeking behaviours are more related to being over-stimulated rather than hypo-stimulated. The final chapter in Part 1, Chapter 6, examined the interrelations between RRB, sensory problems and anxiety. Results showed that both insistence on sameness (IS) and repetitive motor (RM) behaviours were associated with sensory problems. However, only IS behaviours were associated with anxiety while RM behaviours were not. Meditation analyses models showed that sensory sensitivity and anxiety were reinforcing each other in the relationship with IS behaviours.

17 The two chapters in Part 2 examined the frequency and correlates of anxiety in mothers of children with autism. Findings from these two chapters (Chapters 8 and 9) suggested that anxiety is very prevalent among mothers of children with autism with 46% of mothers meeting the cut-off criteria for clinically significant anxiety. For the first time, sensory problems in mothers were examined, and were also found to be very prevalent in mothers, appearing in 60% per cent of mothers. The frequency of sensory features in mothers was correlated with the frequency of sensory features in their children. Further analysis suggested that higher levels of sensory over-sensitivity, higher levels of intolerance of uncertainty, and the dominant use of escape-avoidance coping style contributed to anxiety levels in mothers. Furthermore, it was found that sensory sensitivity in mothers was related to their higher use of escape-avoidant coping strategy and the mediation analysis suggested that sensory sensitivity, escape-avoidance coping and anxiety reinforced each other. The importance of these findings for future theoretical and clinical work is considered in detail in the general discussion chapter (Chapter 10).

18 List of Tables Table Factor Analytic Studies on Repetitive behaviours in autism Table Selection of the sensory processing measures Table Selection of the parents anxiety measure Table Expressive language measure adapted from DISCO Table Questionnaires (about child) included in the Booklet 1 Table Questionnaires (about parent) included in the Booklet 2 Table Health Difficulties in children Table Questionnaire/Interview Evaluation Table Mean RBQ-2 scores at 15, 24 and 72 months Table Mean RBQ-2 scores at 15, 24 and 72 months for each gender Table Response frequencies for each RBQ-2 item Table Factor structure of the RBQ-2 questionnaire Table Correlated item-total correlations for the two factor of the RBQ-2 Table Descriptive statistics Table Relationship between repetitive behaviours and other children s variables Table Anxiety Instruments Evaluation Table Internal Consistency of Spence Children s Anxiety Scale Preschool Version Table Internal Consistency of Spence Children s Anxiety Scale School age Version Table Internal Consistency of Multidimensional Anxiety Scale for Children Table Construct Validity of the Spence Anxiety Scale Table Descriptive statistics for SCAS-P anxiety subscales (N= 32 children) Table Relationship between anxiety levels and other children s variables Table Descriptive statistics for Sensory Modalities and Sensory Quadrants Table Association between sensory quadrants Table Mediation Analysis 1 Table Mediation Analysis 2 Table Seeking as a mediating variable between hypo- and hyper-responsiveness Table Seeking as a mediating variable between hyper- and hypo-responsiveness Table Hyper-sensitivity as a mediator between hypo-sensitivity and seeking Table Hypo-sensitivity as a mediator between hyper-sensitivity and seeking Table Moderators of Sensory Modulation Problems

19 Table Factor structure of RBQ-2 questionnaire items resulting from 2-factor analysis with sensory items excluded Table Correlations between RBQ-2, Sensory Profile Quadrants and Anxiety Table Differences between groups meeting clinical cut-off on the Spence Anxiety Scale Table Mediation Analysis 1 Table Mediation Analysis 2 Table Mediation Analysis 3 Table Mediation Analysis 4 Table Mediation Analysis 5 Table Mediation analysis 6 Table Mediation analysis 7 Table Characteristics of children and their mothers Table Descriptive characteristics of parents and children s variables. Table 8.3. Association between Anxiety in parents and children s variables Table 8.4. Internal Consistency of IRI and WOC-R Table 8.5. Associations between the anxiety in parents, BAP traits and coping styles Table Comparison between anxious and non-anxious group of parents in terms of their coping style and BAP traits Table Associations between coping styles, BAP and chronological age of children Table Effect Sizes Table Internal Consistency of measures Table Descriptive statistics for IUS, HSP and AASP scales Table Performance of parents across four sensory quadrants Table Relationship between sensory atypicalities Table Construct Validity of Highly Sensitive Person Scale and Adult Sensory Profile Table Relationship between anxiety, intolerance of uncertainty and sensory problems Table Differences between anxious and non-anxious groups of mothers Table Predictors of anxiety in mothers Table Predictors of escape-avoidance coping Table Mediation Model 1 Table Mediation Model 2

20 List of Figures Figure Schematic representation of the recruitment process Figure Scree Plot Figure Sensory Profile Scores Across Six Sensory Processing Modalities Figure Sensory Quadrant Scores Figure Graphical representation of the mediated relationship Figure Graphic representation of the Mediation Model 1 Figure Graphic representation of the Mediation Model 2

21 Chapter 1: General Introduction The dread of change and incompleteness seems to be a major factor in the explanation of the monotonous repetitiousness and the resulting limitations in the variety of spontaneous activity. (Kanner, 1943, p246) 1.1. Repetitive Behaviours Leo Kanner (Kanner, 1943) provided a detailed description of various behaviours that are now considered as repetitive behaviours. Behaviours that he observed in his original sample of 11 children included various motor mannerisms such as shaking head from side to side or jumping up and down repeatedly, use of objects in inflexible and highly repetitive manner (e.g., spinning round objects), unusual preoccupation and fascination with ordinary objects such as cardboard boxes and pencils, highly intense interests and very strong insistence that things need to be just so (e.g. insistence that parts of the furniture and other objects need to be arranged in a certain way and becoming distressed if any change was made). Kanner considered these behaviours as an essential feature of then new disorder that has become known as autism and this view has not changed much to this day, exactly 70 years after his seminal paper was published. Since Kanner s original description, various incarnations of international diagnostic criteria required repetitive behaviours to be present in order for ASD diagnosis to be established. The International Classification of Diseases (10 th rev.; ICD-10; WHO, 1993) describe the following four subtypes of Restricted and Repetitive Behaviours (RRBs): (a) encompassing preoccupation with one or more stereotyped and restricted patterns of interest that is abnormal either in intensity or focus; (b) apparently inflexible adherence to specific, non-functional routines or rituals; (c) stereotyped and repetitive motor mannerisms (e.g., hand or finger flapping or twisting or complex whole-body movements); and (d) persistent preoccupation with parts of objects. The latest Diagnostic and Statistical Manual of Mental Disorders (5 th ed.; DSM-V; APA, 2013) lists the following 4 types of repetitive behaviours: (a) extreme attachment to routines and patterns and resistance to changes in routines; (b) repetitive speech or movements; (c) intense and restrictive interests; and (d) difficulty integrating sensory information or strong seeking or avoiding behaviour of sensory stimuli. Although there is a consensus that all these behaviours are linked by rigidity, repetitiveness, invariance and inappropriateness (Turner, 1999), it is still debated whether

22 they should be conceptualized as an uni-dimensional or alternatively as multi-dimensional construct that encompass several related but at the same time distinct behavioural categories. Based on clinical observations and a developmental approach, Prior and Macmillan (1973) and later Turner (1999) argued that RRBs can be classified into lower level behaviours such as dyskinesias, tics, stereotyped movements, repetitive manipulation of objects and selfinjurious behaviours thought to be characteristic for younger and/or lower functioning children and higher level repetitive behaviours such as object attachments, repetitive language, insistence on sameness and circumscribed interests thought to be present in older and more able children. Using a different approach to classification, Lewis & Bodfish (1998) argued that cooccurrence of different repetitive behaviours across various disorders as well as their cooccurrence within individuals suggests that there is likely a common pathophysiology behind various repetitive behaviours and that they can be considered as unidimensional construct (or overall response class ) in this sense. Dysfunction in nigrostriatal dopaminergic pathway was considered as a key problem in the model proposed by Lewis and Bodfish. More recently, Langen and colleagues (Langen, et al., 2010a; b) proposed a more refined neurobiological model that conceptualizes repetitive behaviours as a multidimensional construct with motor stereotypies linked to motor and premotor cortex, rigidity and obsessiveness to cognitive/associative loop and compulsions to anterior cingulateoribtofrontal cortex loop. Over the last 15 years, considerable research effort in the form of factor analytic studies, has focused on the problem of how to describe and classify RRBs. This is not simply a semantic exercise but rather a crucial question and worthwhile enterprise as clear definitions can provide conceptual distinctions that differentiate subtypes of RRB which should facilitate more effective research on origins and outcomes of these behaviours and eventually lead to the development of effective intervention approaches. Therefore this thesis begins with a detailed summary of the classification of RRBs Summary of factor analytic studies that used ADI-R Systematic search of the literature has identified 17 factor analytic studies that have investigated the structure of RRBs in autism. Detailed description of the studies is provided in the table Out of the 17 studies, 11 studies used the Autism Diagnostic Interview-Revised (ADI- R; Lord, Rutter, & Le Couteur, 1994), four studies used the Repetitive Behaviour Scale-

23 Revised (RBS-R; Bodfish et al., 1999), Maudsley Item Sheet (Carcani-Rathwell, Rabe- Hasketh, & Santosh, 2006) and the Repetitive Behaviour Questionnaire (RBQ, Turner, 1995) were used in one study each. A detailed review of RRBs measures will be provided in the Chapter 3. However, before discussing factor analytic studies in more detail, it is appropriate to briefly summarize the above mentioned measures here. The ADI-R is a semi-structured parental interview used in the diagnosis of Autism Spectrum Disorders (ASD). It contains 14 RRBs items with 12 items applicable to children of all age and ability levels and the following 2 items: circumscribed interests and repetitive language which require a certain cognitive ability level. The RBS-R is a 43 item parent questionnaire that asks parents to rate severity and report on the extent to which each item is a problem. RBS-R has the following 6 subscales: stereotyped behaviour, self-injurious behaviour, compulsive behaviour, ritualistic behaviour, sameness behaviour, and restricted behaviour. The RBQ is a 33-item parental questionnaire. RBQ assesses the following four types of RRBs: repetitive movements, sameness behaviour, repetitive use of language and circumscribed interests. The Maudsley Item Sheet is a psychiatric assessment which includes mental state examination and parent interview. It consists of 13 RRBs related items grouped into the following 3 categories: sensory behaviours, repetitive movements and cognitive rigidity symptoms. Out of the 11 factor analytic studies that used the ADI-R, eight studies identified 2 factor solution, two studies identified 3 factor solution and one study identified only one factor. In studies that found a two factor solution, the two factors were generally labelled as repetitive sensory-motor (RSM) behaviours and insistence on sameness (IS) behaviours. The third factor was labelled as Restricted Interests/Preoccupations. As can be seen from Table 1.1, the findings regarding RSMB factor were quite consistent across the studies. For example, in the examination of the RRBs of 292 individuals with autism (age range: 3-21 years), Cuccaro et al. (2003) used 12 ADI-R items and identified a RSMB factor consisting of the following items: hand and finger mannerisms, unusual sensory interests, repetitive use of objects or parts of objects, complex mannerisms or stereotyped body movements and rocking. With the exception of the rocking item, the RSMB factor identified by Cuccaro and colleagues contained identical items as the RSMB factor identified by Honey et al. (2006), Szatmari et al. (2006) and very similar items found by other 6 ADI-R factor analytic studies. The situation was more complicated regarding the IS factor. Although the following items: difficulties with minor changes in routine or personal environment, resistance to trivial

24 changes in the environment and compulsions/rituals, consistently loaded onto an IS factor across all 9 studies, other items such as the unusual attachment to objects, unusual preoccupations and restricted interests showed inconsistent loading. More specifically, these items either loaded onto the IS factor (Mooney et al., 2009), did not load onto the IS factor and were not included in the final factor solution (Cuccaro et al., 2003) or, together with some other IS items constituted Restricted Interests/Preoccupations factor as in Lam, Bodfish & Piven (2008) and Honey et al. (2006) studies. Two factors that resemble RSMB and IS factor identified in ADI-R factor analytic studies were also found by Carcani-Rathwell et al. (2006) who used Maudsley Item Sheet and by Honey et al. (2012) who used RBQ. Two studies that used the RBS-R to investigate the structure of repetitive behaviours identified more complex factor solutions than studies reviewed above. In a sample of 307 individuals with autism, Lam et al. (2007) found evidence for the existence of the following 5 factors: Rituals/Sameness, Self-injurious Behaviour, Stereotypic Behaviour, Compulsive Behaviour and Restricted Interests. A similar factor structure of RBS-R to that identified by Lam et al. (2007) was also reported by Bishop et al. (2013) in a large sample of 1825 individuals with autism aged 4-18 years. As can be seen from Table 1. 1., the age range of individuals with autism included in this study was very wide, ranging from 3-48 years. This limitation was addressed by Mirenda et al. (2010) who explored the structure of RRBs in a population 287 children aged 2-5 years. Factors identified by Mirenda et al. were very similar to the ones identified in the previous study by Lam and colleagues. The third factor analytic study that used the RBS-R study (Georgieades et al. 2010) found evidence for two factors in a sample of 205 individuals with autism aged 2-48 years. These findings are more in line with ADI-R and other studies reviewed above. Georgiades and colleagues found that all items from the following four original RBS-R subscales: Compulsive, Ritualistic, Sameness, and Restricted behaviours loaded onto a single factor resembling IS behaviours and that all items from the original RBS-R Stereotypy and Self-Injurious subscales loaded onto the factor that somewhat resembled RSM behaviours. The studies reviewed above differed in the characteristics of participants and measures used. As can be seen from Table 1.1, some studies have examined RRBs structure in toddlers and very young children (Honey et al., 2008; Mooney et al., 2009), however, samples in the rest of the studies consisted of individuals with very wide chronological age range and also variable developmental level. Equally important is the question of the specific measure that was used. I will discuss this issue in depth in Chapter 3. However, it is important to highlight here that the type of measure that is used to assess RRBs sets the

25 boundaries for the phenomenon under focus and necessarily influences the conceptualization of RRBs. For example, although the ADI-R is a dominant measure of RRBs, it is limited in two ways. Firstly, it has been highlighted by several authors that because the ADI-R is used to diagnose children with autism, its use in research raises the issue of circularity (Lam et al., 2007; Leekam et al., 2011; Honey et al., 2012). Secondly, in addition to the issue of circularity the ADI-R has only 12 RRB items that are applicable to children of all ages and developmental levels and it has been reported that it undersamples RRBs (Lam et al., 2007; Honey et al., 2012). Further to this, as can be seen from Table 1. 1., studies were inconsistent in terms of the numbers of ADI-R items that are included in the analysis. Although the RBS- R is a measure that assesses a much wider range of behaviours when compared to ADI-R, most of the items are appropriate to individuals with lower developmental level only. In addition to this, the RBS-R includes self-injurious behaviours (SIBs) which are considered by a number of authors to be conceptually different than repetitive behaviours (a comprehensive review of this issue is provided in Matson, Dempsey, & Fodstad, 2009). In summary, as Honey et al. (2012) pointed out, it can be argued that it is difficult to achieve a consensus regarding the conceptualization of RRBs in the absence of a good measure. The variability in the samples examined and in RRBs measures, necessarily limits conclusions that can be drawn from the reviewed literature. The reviewed studies seem to be relatively consistent regarding the existence and content of RSM factor that resembles behaviours that Turner (1999) termed as low-level behaviours. Findings regarding the group of behaviours that Turner considered as higher level behaviours are much less consistent. More precisely, these behaviours form either one or two categories of behaviours and specific behaviours that form these categories are highly variable between studies. However, in general, results seem to lend support to the categories proposed by Turner (1999) and listed in the current diagnostic criteria.

26 Table Factor Analytic Studies on Repetitive behaviours in autism Author (year) Cuccaro al. (2003) et Autism Group (Number of participants, age, functioning/diagnostic group) N= 292 individuals with autism, CA range: 3-21 years. RRBs Measure ADI-R (12 items) Factors Sensory-Motor Insistence on Sameness Restricted Interests hand and finger mannerisms, unusual sensory interests, repetitive use of objects or parts of objects, other complex mannerisms or stereotyped body movements and rocking difficulties with minor changes in routine or personal environment, resistance to trivial changes in the environment and compulsions/rituals Items not included in the final model unusual preoccupations, unusual attachment to objects, idiosyncratic negative responses, sensitivity to noise Shao et al. (2003) N= 221 individuals with autism, CA range: 3-21 yr. ADI-R (13 items) hand & finger mannerisms, repetitive use of objects, unusual sensory interests, complex mannerisms, rocking, unusual preoccupations, unusual attachment to objects difficulties with minor changes in personal routine or environment, resistance to trivial changes in the environment, compulsions/rituals circumscribed interests, sensitivity to noise, idiosyncratic negative responses Tadevosyan- Leyfer et al. (2003) N= 292 individuals with autism, age range: 2-47yr ADI-R (98 items in total, number of RRB items was not specified) Only one Factor (Compulsions): Stereotyped utterances, Unusual preoccupations, Compulsions/rituals, Resistance to trivial changes in the environment, Unusual attachment to objects. Repetitive motor behaviours didn t load to compulsions factor but loaded with items related to Social Interaction.

27 Author (year) Carcani- Rathwell et al., 2005 Bishop, Richler & Lord (2006) Autism Group (Number of participants, age, functioning/diagnostic group) N= 319 individuals with PDD (183 with PDD+ MD and 119 with PDD without MD). Mean CA= 7.7 years (SD= 4.1) N=830 children with autism (N=560 autism, N=268 PDD- NOS, 2 AS). Mean CA 4.8 yr (range: ). RRBs Measure Maudsley Item Sheet ADI-R (11 items) Factors Sensory-Motor Insistence on Sameness Restricted Interests thumb sucking, tongue sucking, rocking, masturbation, nail biting, scratching, head banging, whirling, flapping, twisting of hands, complex mannerisms repetitive use of objects, unusual sensory interests, complex mannerisms, hand & finger mannerisms, unusual preoccupations circumscribed interests and preoccupations, Resistance to change, Repetitive lining up difficulties with minor changes in personal routine or environment, resistance to trivial changes in the environment, compulsions/rituals, idiosyncratic negative responses, sensitivity to noise Items not included in the final model unusual attachment to objects. Szatmari et al. (2006) N= 339 individuals with autism, mean CA: 8.4 yr (SD = 5.5) months. Mean IQ (Leiter) = 65.7 (SD = 28.7). ADI-R (11 items) difficulties with minor changes in personal routine or environment, resistance to trivial changes in the environment, compulsions/rituals hand & finger mannerisms, repetitive use of objects, unusual sensory interests, complex mannerisms, rocking circumscribed interests, unusual preoccupations, unusual attachment to objects didn t load on any of the factors.

28 Author (year) Lam et al. (2007) Richler et al. (2007) Autism Group (Number of participants, age, functioning/diagnostic group) N= 307 individuals with autism. Mean CA= years (age range: 3-48) N= 165 AUTISM (CA < 3yr), N= 44 DD ( yr) and N= 65 TD children (CA < 3yr). Children were assessed when they were 2, 3, 5 and 9yr old. RRBs Measure RBS-R ADI-R (12 items) Factors Sensory-Motor Insistence on Sameness Restricted Interests Factor 1(Rituals/Sameness): placement of objects, no new places, no interruption, walks certain way, sits certain place, appearance/behaviour of others, uses certain door, videotapes, difficult transitions, insists on routine, insists on time; Factor 2 (Self-injurious Behaviour): hits body, hits against surface, hits object, bites self, pulls hair/skin, rubs/scratches, inserts finger/object, picks skin; Factor 3 (Stereotypic Behaviour): body movements, head movements, finger movements, locomotion, object usage, sensory; Factor 4 (Compulsive Behaviour): ordering, completeness, washing, checking, counting, hoarding; Factor 5 (Restricted Interests): preoccupation with subject, attached to object, preoccupied with parts of object, preoccupation with movement. repetitive use of objects, unusual sensory interests, hand/finger mannerisms, and other complex mannerisms compulsions and rituals, difficulties with changes in routine, and resistance to trivial changes in the environment Items not included in the final model unusual preoccupations, unusual attachment to objects, sensitivity to noise, and abnormal/idiosyncratic response to sensory stimuli Honey et al. (2008) N= 104 children with autism or language disorders, CA range: 2-4 yr. ADI-R (12 items) hand & finger mannerisms, repetitive use of objects, unusual sensory interests, complex mannerisms, unusual fears, self injury difficulties with minor changes in personal routine or environment, resistance changes in environment, compulsions/rituals, unusual fears, idiosyncratic negative responses unusual preoccupations, unusual attachment to objects

29 Author (year) Lam, Bodfish & Piven (2008) Mooney et al. (2009) Georgiades et al. (2010) Autism Group (Number of participants, age, functioning/diagnostic group) N= 316 Individuals with autism. Mean CA=9.02 yr (range ). Mean IQ=69.5 (range ). N= 137 DD children with PDD and N= 61 DD children without PDD, all with CA yr. N= 205 individuals with autism. Mean CA= 11.5 years (range: 2-48) RRBs Measure ADI-R (10 items) ADI-R (12 items) RBS-R Factors Sensory-Motor Insistence on Sameness Restricted Interests repetitive use of objects, hand and finger mannerisms, and other complex mannerisms/stereotyped body movements items hand & finger mannerisms, repetitive use of objects, complex mannerisms difficulties with minor changes in personal routine and environment, resistance to trivial changes in the environment, and compulsions and rituals difficulties with minor changes in personal routine or environment, resistance to trivial changes in the environment, compulsions/rituals, unusual attachment to objects circumscribed interests, unusual preoccupations and unusual attachment to objects Factor I (Compulsive Ritualistic Sameness Restricted Behaviours; CRSRB): all items from the original RBS-R Compulsive, Ritualistic, Sameness, and Restricted subscales. Factor II (Stereotyped Self-Injurious Behaviours; SSIB): all items from the original RBS-R Stereotypy and Self-Injurious subscales. Items not included in the final model Rocking idiosyncratic negative responses, sensitivity to noise, unusual sensory interests, unusual preoccupations, selfinjury Mirenda et al. (2010) N= 287 children with autism, mean CA3.4 yr, range: RBS-R 3 and 5 factor RRBs models were the preferable models. Model III comprised of Factor I: Compulsive Ritualistic Sameness Behaviours (CRSB); Factor II: Self Injurious Behaviours (SIB), Factor III: Restricted Stereotyped behaviours (RSB), items 1 6 and Model V comprised of Factor I: Stereotyped Behaviours; Factor II: Self Injurious Behaviours (SIB); Factor III: Compulsive Behaviours (COMP); Factor IV: Ritualistic Sameness Behaviours (RITUAL/SAME) and Factor V: Restricted Behaviours (RESTR).

30 Author (year) Richler et al. (2010) Autism Group (Number of participants, age, functioning/diagnostic group) N= 192 children with autism who were referred for a diagnosis when they were under the age 3 and followed up at the age of 3, 5 and 9. RRBs Measure ADI-R Factors Sensory-Motor Insistence on Sameness Restricted Interests repetitive use of objects, unusual sensory interests, hand/finger mannerisms, and other complex mannerisms compulsions and rituals, difficulties with changes in routine, and resistance to trivial changes in the environment Items not included in the final model unusual preoccupations, unusual attachment to objects, sensitivity to noise, and abnormal/idiosyncratic response to sensory stimuli Honey et al. (2012) N= 180 individuals with autism. Mean CA= months (age range: ). N= 40 children had FSIQ< 70 RBQ spin around, rock backward, forwards or side to side, pace or move around, hand and finger mannerisms, body movements Insistence on Sameness/Restricted Interests Factor: insists things in the house stay the same, insists items put out or stored in the same way, plays the same music, game or video, insists on using same objects or items in specific situations, insists wearing the same clothes or refuse new clothes, certain items of clothing always worn, insists on eating same foods, insists on moving or travelling by same route, react to changes to surroundings at home, insists on same routines, rituals for everyday activities, rituals linked to particular occasions and places, negative reaction to changes in daily routine says the same things or make same noises, talks about same topic, operate light switches, taps toilet flush, arrange toys or other items,

31 Author (year) Bishop et al., 2013 Bishop et al., 2013 Autism Group (Number of participants, age, functioning/diagnostic group) N= 1825 Individuals with autism. Mean CA= 8.9 years (range: 4-18 years, SD: 3.5). Mean VIQ: 79.5 (range: 5-167, SD= 30.2), mean NVIQ: 86.5 (range: 9-161, SD= 25.1), mean FSIQ: 83 (range: 7-167, SD: 26.9) Same as above RRBs Measure ADI-R RBS-R Unusual preoccupations, repetitive use of objects, hand and finger mannerisms, complex mannerisms, unusual sensory interests Factors Compulsions/rituals, sensitivity to noise, abnormal/idiosyncratic response, difficulties with change, resistance to change, circumscribed interests Factor 1(Sensory motor): ): body movements, head movements, finger movements, locomotion, object usage, sensory, fascination with movements; Factor 2 (Self-injurious Behaviour): hits body, hits against surface, hits object, bites self, pulls hair/skin, rubs/scratches, inserts finger/object, picks skin; Factor 3 (Compulsive Behaviour): ordering, completeness, washing, checking, counting, hoarding, repeating; Factor 4 (Ritualistic/Sameness Behaviours): ordering, completeness, washing, checking, counting, hoarding; Factor 5 (Restricted Interests): preoccupation with subject, attached to object, preoccupied with parts of object, preoccupation with movement. Self-injurious behaviours, Unusual attachment to objects

32 How do repetitive behaviours in autism change over time? Another way to investigate whether proposed distinctive and yet related categories of RRBs are valid is to look at whether there is a difference in the way that these behaviours are related to other characteristics of individuals with autism, in particular their chronological age and developmental level. This literature will be reviewed below (studies are also summarized in Table 1, Appendix 1). The current empirical findings are mixed regarding the association between age and the number and severity of RRBs in autism. In a retrospective study of changes in the diagnostic triad, Piven, Harper, Palmer & Arndt (1996) used the ADI-R to examine the current behaviours of 38 high functioning adults and adolescents with autism (mean age= 17.6 years). These behaviours were then compared with the behaviours that these individuals showed at 5 years of age (according to their parents retrospective reports). It was found that repetitive behaviours improved in only 50% of the sample. Similar methodology was also employed by Fecteau, Mottron, Berthiaume and Burack (2003) who examined changes in the triad features of 28 children and adults with autism. Fecteau and colleagues identified more positive changes in repetitive behaviours over time, although not as great as those seen for social and communication impairments. They found that the greatest improvements were in the repetitive use of objects. More precisely, of 17 children who were reported to engage in the repetitive use of objects at 4 to 5 years of age, 76.5% appeared to show a decrease in these behaviours with age. Also, with the exception of verbal rituals, no behaviours got worse over time. The findings above were based on retrospective report of changes. Several longitudinal studies have also been carried out. For example, longitudinal study by Moore & Goodson (2003) examined twenty children at 2 years and again at 4 to 5 years using the ADI- R. While there was little change in social interaction and communication skills over time, repetitive behaviours became more apparent. Changes in the ADI-R scores of 29 children with autism were examined by Charman, Taylor, Drew, Cockerill, Brown and Baird (2005). This study showed that repetitive behaviours increased between 3 years and 4 to 5 years before decreasing from 4-5 years to 7 years. In a study that examined the way in which repetitive behaviours changed from 2 to 9 years of age in a sample of 172 children with autism and PDD-NOS, Lord, Risi, DiLavore, Thurm & Pickles (2006) found that repetitive behaviours increased with age. Finally, using the Autism Diagnostic Observation Schedule

33 33 (ADOS; Lord et al., 2002), Kim and Lord (2010), did not find that chronological age was associated with the severity of repetitive behaviours in a sample of 121 children with autism and 71 children with PDD-NOS divided into the following 6 age cohorts (younger than 18 months, 19 24; 25 30; 31 36; 37 42; and months). The reasons for reported inconsistencies in terms of the relationship between chronological age and repetitive behaviours are two-fold. Firstly, as studies by Evans, Gray and Leckman (1999) and Berkson and Tupa (2000) show, developmental level is negatively associated with repetitive behaviours in severely delayed populations. These findings, together with the fact that although chronological and developmental age are parallel in typical development, developmental delay is common in children with autism, suggest that it is necessary to consider the developmental level of children with autism when examining presence and changes in their behaviours. Secondly, as highlighted in the section on the structure of repetitive behaviours, it has been suggested that different types or classes of repetitive behaviours are differently associated with chronological and developmental age. The evidence for a relation between repetitive behaviour and developmental ability will be reviewed below. In addition to the question of how repetitive behaviours change with chronological age, Piven et al. (1996) also examined differences based on cognitive level. When the autism group was split according to IQ, it was found that the greatest decrease in repetitive behaviours with age was for a group of Individuals with autism with an IQ over 70 suggesting that the greater the cognitive capacity of an individual, the greater their decrease in repetitive behaviours with age. Gabriels, Cuccaro, Hill, Ivers & Goldson (2005), like Piven et al., divided individuals with autism into high nonverbal mental age group (NVIQ> 97) and low nonverbal mental age group (NVIQ< 56). Using RBS-R to examine differences in repetitive behaviours between groups, they found that total RRBs scores were significantly higher in lower NVIQ group. After adjusted for multiple comparisons it was revealed that the groups differed significantly on only the Sameness scale (i.e., difficulties with change) which was more severe in the Low NVIQ group. A study by Bishop, Richler & Lord (2006) used the ADI-R in order to examine the effects of non-verbal IQ and chronological age on repetitive behaviours in 830 children with autism (age range: 6 to 12 years). This study examined the prevalence and severity of thirteen repetitive behaviours, that is, whether behaviours were present or not and to what extent they interfere with daily life. With respect to age, analysis revealed significant improvements in the prevalence of some ADI-R

34 34 repetitive behaviour items including the following: sensitivity to noise, circumscribed interests, difficulties with changes in routine and resistance to trivial changes in the environment. On the other hand, the repetitive use of objects and unusual sensory interests were found to increase significantly. With respect to non-verbal IQ, lower IQ indicated a larger number of behaviours including the repetitive use of objects while higher IQ levels indicated the higher presence of circumscribed interests. Esbensen et al. (2009) used RBS-R to examine age related differences in 700 individuals with autism aged 2 to 62 years. They found that older individuals had significantly lower RBS-R total scores as well as lower scores for all subscales. Furthermore, individuals with autism and comorbid ID had significantly more stereotyped movements and SIB than autistic individuals without the comorbid ID. ID was not significantly correlated with expression of ritualistic/sameness behaviours, compulsions and restricted interests. More recently, Richler, Huerta, Bishop & Lord (2010) used the ADI-R to track longitudinal change in RRBs in children with ASD over a 7 year period (RRBs were assessed when children were 2, 3, 5, and 9 years of age). Results showed that RSM behaviours remained relatively stable over a 7 year period but RSM behaviours did decrease in children with higher NVIQs. Unlike RSM behaviours, IS behaviours, gradually (moderately) increased over time. In addition to older chronological age, higher IS scores were also associated with milder social and communication impairments. Possible reason for this finding was the fact that age/iq ration was too low. The findings reviewed above and summarized in Table 1 (Appendix 1) highlight the importance of considering the role of developmental level in addition to the role of chronological age and suggest that different types of RRBs are differently associated with chronological and developmental age. The majority of studies seem to support Turner s (1999) conceptualization of RSM behaviours as lower level behaviours as they are generally found in younger children and individuals with lower developmental level. Conceptualization of IS behaviours as higher order is less straightforward. Several studies failed to find association between this factor and developmental level. However, it is important to highlight that these findings need to be considered against what we know about the structure and change of repetitive behaviours during typical development. A brief summary of repetitive behaviours in typical development will be reviewed below.

35 Repetitive Behaviours in Typical Development Repetitive behaviours are seen throughout childhood in the normative population and are considered to be an important feature of development. A variety of rhythmic and stereotyped motor and sensory behaviours are very prevalent during the first year of life (Thelen, 1979; 1981). It has bee suggested that these behaviours serve adaptive purposes such as neuromuscular and motor skills development (Wolff, 1968; Thelen, 1979) and general central nervous system maturation (Sprague & Newell, 1996) and begin to reduce after the child's first birthday. For example, Leekam et al. (2007) used the Repetitive Behaviour Questionnare-2 (RBQ-2) to explore repetitive sensory and motor behaviours in 675 two-year old TD children and found that, at that age, these behaviours were common, with each of RBQ-2 items being endorsed by between 18 and 30% of children. Arnott et al. (2010) explored the frequency of RRBs in the same sample of children as Leekam and colleagues (children were 15 month old) and found that motor behaviours were more frequent than when children were 24 months old, with some items such as repetitively fiddling with toys, being endorsed by 60% of children. At the same time, when sensory and motor repetitive behaviours such as kicking and banging start to decrease, rigid types of behaviours such as a need for sameness start to increase. Gessell and colleagues (1928; 1974) observed that between 2 to 3 years of age, typically developing children show compulsive like patterns of behaviours such as insistence on sameness, repetitive and ritualised behaviours, rigidity in likes and dislikes and acute sensory perception for minute details. In a parental-report study of 8 to 72 month-old typically developing children, Evans, Leckman, Carter, Reznick, Henshaw, King & Pauls (1997) used the Childhood Routines Inventory (CRI, Evans et al., 1997) to examine the developmental trajectory of two components of 'compulsive-like' behaviours. These were 'just right' behaviours which included higher level behaviours (such as a preference for the same daily routine, repetition of certain actions over and over, acting out the same sequence over and over in pretend play), and 'repetitive behaviours & insistence on sameness' (including items such as: preference for the same household schedule or routine every day, preference for certain foods). With respect to 'just right' behaviours, they occurred significantly more in children aged 24 to 48 months than in children aged 12 months and 72 months who displayed similar lower levels of these behaviours. Children younger than 12 months demonstrated the lowest levels of 'just right' behaviours. With respect to 'repetitive behaviours and insistence on sameness', these behaviours were present to a similar degree as

36 36 'just right' behaviour, although these behaviours emerged earlier and had a steeper increase over time. In a longitudinal study using the RBQ-2 described above, Leekam et al. (2007) found that at the age of two, items: 'fascination with specific objects' and 'carry special objects around' were reported as 'marked' behaviours in more than 30% of the sample and as 'mild' and 'occasional' in 40% and 28% of their sample of 675 TD children respectively. The findings reviewed above suggest that RSM and IS behaviours have somewhat distinctive developmental trajectories. RSM behaviours seem to be apparent very early in development and start to quickly reduce in the frequency and intensity from around months while IS behaviours start to increase, about the same time (15-18 months) and are prominent up to around 4 years when they start to decrease. It is not clear whether RSM behaviours are precursors or earlier developmental forms that are then replaced by the IS behaviours or whether, these two behaviours represent independent classes of behaviours with separate genetic and neurophysiological make-up. Exploring this question is of crucial importance for better understanding of RRBs in autism as well as in other disorders. Repetitive behaviours are not only present during early typical childhood and are not only found in individuals with autism but also occur in a wide variety of neurodevelopmental conditions and individuals with developmental disabilities, psychiatric disorders (for example schizophrenia, obsessive-compulsive disorder), and neurological conditions such as Tourette s Syndrome for example. It is important to determine whether any of repetitive behaviours are specific to autism. The literature on repetitive behaviours in other conditions will be very briefly summarized below and the question of distinctiveness of repetitive behaviours between autism and other clinical groups will be discussed Repetitive Behaviours in other neurodevelopmental and psychiatric conditions Stereotyped hand and body mannerisms are common in individuals with mental retardation, individuals diagnosed with schizophrenia, Tourette s syndrome, attention deficit hyperactivity disorder (ADHD), Alzheimer s Disease and other forms of dementias, in individuals with sensory impairments such as loss of hearing and vision (Bodfish, Symons, Parker, & Lewis, 2000; Mahone, Bridges, Prahme, & Singer, 2004). Rigid behaviours in the form of obsessions and compulsions and behaviours with insistence on sameness component are a defining feature of OCD as well as in anxiety spectrum conditions (Robinson, 1998). It has been found that in some instances certain repetitive behaviours can be considered

37 37 characteristic of certain syndrome groups. For example, lick and flip and self-hug behaviours are frequently described in Smith Magenis syndrome (Finucane, Dirrigl, & Simon, 2001). Attachment to objects is very prevalent in Cri du Chat syndrome (Cornish & Pigram 1996) and skin picking is present in majority of individuals with Prader-Willi Syndrome (PWS; Dykens, Leckman, & Cassidy, 1996; Dykens, Cassidy, & King, 1999; Butler et al. 2004; Torrado et al. 2006). The question naturally arises, what distinguishes repetitive behaviours seen in individuals with autism from repetitive behaviours observed in conditions reviewed above? Distinctiveness of repetitive behaviours in autism when compared with other clinical groups Turner (1999) suggested that in addition to being associated to younger chronological age and lower developmental age, repetitive sensory and motor behaviours do not distinguish the population with autism from other clinical groups where cognitive impairments are present. Findings from several comparative studies seem to confirm this hypothesis (Greaves, Prince, Evans, & Charman, 2006; Moss & Oliver, 2009; Flores et al., 2011). Evidence that insistence on sameness type of behaviours is more specific to autism are mixed. These types of behaviours are, as mentioned above, certainly present in other conditions and are even considered as defining characteristics of some disorder such as OCD for example. A study by Greaves et al. (2006) used the CRI to examine RRBs in 89 children with autism and 80 children with PWS and didn t find differences between children with PWS and children with autism on either total CRI scores or just right and repetitive behaviour subscale scores. However, more recent research by Flores et al. (2011) found that RRBs were significantly less frequent and less severe in the PWS sample when compared with the autistic sample (RBS-R was used). The use of different measures could have been the contributing factor to the inconsistent findings. Two studies compared the repetitive behaviours of individuals with autism and OCD. Zandt, Prior, & Kyrios (2007) used the RBQ and Children s Yale Brown Obsessive-Compulsive Scale (CY-BOCS; Scahill et al., 1997) to examine repetitive behaviours in 19 children with autism and CA and IQ matched children with OCD (N=17). They found that although both groups showed similar levels of IS behaviours, children with OCD had more compulsions and obsessions than children with autism. There were also differences in terms of how age influenced the expression of repetitive behaviours, with younger children with OCD showing more IS behaviours while for children with autism, age

38 38 was not significantly associated with the expression of IS behaviours, repetitive movements, compulsions or obsessions. Similar findings to the ones reported by Zandt et al. (2007) in children were also found when comparing adult ASD and OCD populations (McDougle et al., 1995). The picture that seems to be emerging from reviewed studies is that it is the frequency of RRBs rather than their form or pattern that distinguishes autism from other clinical groups. In other words, while more specific behaviours can be seen in other clinical group, autistic group tends to express a wide range of RRBs at the same time Distinctiveness of RRBs as a core impairment or the relationship between repetitive behaviours and other core autism symptoms Traditionally, it is assumed that repetitive behaviours are related to the impairments in reciprocal social interaction and deficits in communication. However, this assumption has been questioned by several researchers (Constantino, Gruber, Davis, Hayes, Passanante, & Przybeck, 2004; Ronald, Happe & Plomin, 2005; Happe, Ronald, & Plomin, 2006; Happe & Ronald, 2008). For example, Constantino et al. (2004), using both ADI-R and Social Responsiveness scale across different psychiatric conditions found that a single, unitary factor solution, rather than three factor, provided the best description of their data. On the other hand, Happe et al. (2006) and Happe and Ronald (2008) argued for the fractionation of social and non-social (RRB) symptoms of autism. They suggested that social and nonsocial impairments have distinctive genetic etiologies and that their association in children with autism is coincidental. This suggestion was supported to an extent by findings from Ronald et al. (2005) who found only modest correlation between RRBs, social and communication problems. Furthermore, evidence that pointed out to the independent, at least to certain extent, genetic influence on social and non-social symptoms were found. Several studies have explored the relationship between RRBs and social/communication problems (Charman et al., 2005: Gabriels et al., 2005; Szatmari et al., 2006; Lam, Bodfish & Piven, 2008; Paul, Chawarska, Cicchetti, & Volkmar, 2008; Watt Wetherby, Barber, & Morgan, 2008; Richler et al., 2010; Ray-Subramanian & Weismer, 2012) and in general, found that higher RRBs scores were associated with more social/communication problems. For example, Ray-Subramanian & Weismer (2012) examined the association between repetitive behaviours, and language and nonverbal

39 39 cognitive skills in a sample of 115 children with autism. All the variables were assessed at two age points: at the ages of 2 and 3. They found that, by the age of 3, higher RRBs scores were correlated with lower levels of receptive and expressive language, and NVIQ. It was also reported that the increase in language levels between ages 2 and 3 predicted decrease in RRBs. A study by Paul et al. (2008) looked at the language outcomes in 37 children with autism. Children were assessed at two time points-when they were months old and 12 months later using ADI-R and VABS. It was found that lower levels of repetitive behaviours during the initial assessment (together with higher receptive language VABS scores) were significant predictors of VABS expressive language scores 12 months later. Studies that looked at subtypes of repetitive behaviours found that they were differently associated with communication and social abilities For example, Lam, Bodfish & Piven (2008) used ADI-R in a sample of 316 children with autism (mean age= 9 years) to look at the relationship between different RRBs subtypes and social/communication deficits. They found that both higher repetitive motor behaviours and insistence on sameness were significantly associated with more social/communication problems. No association between circumscribed interests and social/communication deficits was found. Also, Szatmari et al. (2006) found that while repetitive sensory-motor behaviours (ADI-R) were negatively associated with all VABS subscales and ADI-R communication scores, insistence on sameness behaviours showed positive association with VABS and ADI-R communication scores Neurobiology of Repetitive Behaviours Basal ganglia are the structures most often connected with repetitive behaviours in autism although the structural and functional findings have been somewhat inconsistent (Langen et al., 2010). Hollander et al. (2005) found a 10% larger right caudate volume in a group of 17 individuals with autism (mean age= 29.4 years, SD= 9.08) and a lack of asymmetry between left and right caudate when compared with typically developed controls. The insistence on sameness repetitive behaviour scores (measured with ADI-R) showed a positive correlation with the right caudate and putamen volumes. However, Sears et al. (1999) found no differences in the volume of caudate (and no evidence for symmetry reversal), putamen and globus pallidus between 35 individuals with autism (age range: years) and typically developed controls. Caudate volume was negatively associated with the following ADI-R items: complex mannerisms, compulsions/rituals, and difficulties with

40 40 minor changes in routine were significantly negatively correlated with caudate volume (ADI was used). Other structures, such as anterior cingulate cortex (ACC) and amygdala, have also been implicated in the neurobiology of repetitive behaviours. For example, Thakkar et al. (2008) examined the association between ACC structure and RRBs. They found reduced fractional anisotropy in white matter underlying dorsal and rostral ACC in 12 adults with autism. Lower fractional anisotropy was significantly associated with higher repetitive behaviour scores (ADI-R). Furthermore, individuals with autism made more antisaccade errors, showed increased rostral ACC activation to both correct and error responses, and increased activation in rostral ACC to correct responses was positively associated with ADI- R repetitive behaviour scores. Although amygdala abnormalities have been traditionally associated with the social impairments (Baron-Cohen et al. 2000; Ashwin et al., 2006), Amaral and Corbett (2003) suggested that it might contribute to anxiety and repetitive behaviours. Indeed, Dziobek et al. (2006) found that higher imposition of routines and rituals (ADI-R) was associated with smaller amygdala volume in 17 individuals with Asperger syndrome (mean age= 41.4). However, although interesting, this finding needs to be replicated. It is important to emphasize that both methodological differences such as manual anatomic tracing versus automated MRI morphometry, as well as differences between participants, may have contributed to the above mentioned inconsistencies in findings between the studies. For example, one particularly interesting issue might be the age difference between samples. Langen et al. (2009) investigated the structural brain development of caudate nucleus in 99 individuals with high functioning autism (mean age= years, SD= 4.45) and found that while caudate volume in the autism group increased with age it decreased in the control group. In summary, it can be argued that in order to gain better understanding of the neurobiology of repetitive behaviours, it is necessary to explore these behaviours across different conditions and also to go beyond separate brain structure and investigate neural circuits and the role of their dysfunction in repetitive behaviours. This approach was advocated by Langen and colleagues in two recent review articles (Langen et al., 2010a, b). They proposed that motor stereotypies can be linked to the dysfunction in motor and premotor cortex, rigidity and obsessiveness to cognitive/associative loop and compulsions to

41 41 anterior cingulate-oribtofrontal cortex loop. Future experimental work will need to test and further refine this proposal Summary Despite being recognized as a core autism feature from the original description by Kanner and Asperger to the latest instalments of the current diagnostic criteria, there is still much to be learned about the definition and structure of RRBs in autism, their cause and potential to change, their relationship with other core autism characteristics. Although several new measurements of RRBs have been developed, as could be seen from the brief overview of the literature presented in this section, the majority of research is still dominated by the ADI-R whose use in the research on RRBs raises the issue of circularity and undersampling of RRBs. As has been emphasized in this section, characteristics of the measure that is used to assess RRBs (or any other behaviour or phenomenon for that matter) sets the boundaries and necessarily influences the conceptualization of RRBs. Having this in mind, one of the main aims of the first empirical chapter on RRBs (Chapter 3) will be to provide a more detailed review of the RRBs measures and how the RRBs measure used in this thesis was selected. This chapter will then present results of a factor analysis and exploration of the psychometric properties of the selected RRBs measure. The second focus of Chapter 3 will be to look at the developmental trajectory of two hypothesised classes of RRBs: RSM and IS in typical development. As has been discussed above, a better understanding of the change in RRBs in typical development is of crucial importance for the better understanding of these behaviours in autism. Finally, Chapter 3 will also address the issue of how different types of RRBs are related to the core autism features in children and adolescents with autism. In subsequent chapters the focus moves to the issue of potential factors for the maintenance of RRBs in children with autism. More precisely, I focus on the possibility that anxiety and sensory problems serve as important triggers for RRBs.

42 Anxiety and Autism The presence of anxiety in individuals with autism was noted in the original descriptions by Leo Kanner (Kanner, 1943) and Hans Asperger (Asperger, 1944). Fear of unusual objects and situations such as sound of egg beater, vacuum cleaner, playgrounds and various toys were frequent in children that both Kanner and Asperger described. Kanner noted that children with autism displayed an anxiously obsessive desire for the maintenance of sameness and further speculated that some of the core autism features might be, at least in part, driven by high levels of anxiety. Since these original observations, numerous clinical and autobiographical accounts demonstrated the negative impact of anxiety in autism. Vivid descriptions of the negative impact of these difficulties were provided by Joliffe, Landsdown and Robinson (1992), with one description stating: reality to an autistic person is a confusing, interacting mass of events, people, places, sounds and sights. There seem to be no clear boundaries, order or meaning to anything. A large part of my life is spent trying to work out the pattern behind everything. Set routines, times, particular routes and rituals all help to get order into an unbearably chaotic life. Trying to keep everything the same reduces some of the terrible fear. The problem of anxiety in autism has been largely neglected until relatively recently despite the clear indications of its pervasiveness and debilitating influence on the lives of Individuals with autism. This is illustrated by the fact that the first systematic review of anxiety in autism was published in 2009 (White, Oswald, Ollendick, & Scahill, 2009). Discussion in this section will focus on the prevalence, moderators and current theories of anxiety in autism, briefly touching on the issue of assessment. Assessment of anxiety in populations with autism will be discussed in more depth in Chapter Prevalence of anxiety in autism The existing research studies show that individuals with autism exhibit higher levels of anxiety than typically developed individuals (Kim, Szatmari, Bryson, Streiner, & Wilson, 2000; Bellini, 2004; Bradley, Summers, Wood, & Bryson, 2004; Gadow, DeVincent, Pomeroy, & Azizian, 2004; Weisbrot, Gadow, DeVinvent, & Pomeroy, 2004; Thede & Coolidge, 2007; Lopata et al., 2010), individuals from other clinical groups, such as Down

43 43 Syndrome (Evans, Canavera, Klinepeter, Taga, & Maccubbin, 2005), Conduct Disorder (Green, Gilchrist, & Cox, 2000), Specific Language Impairment (Gillott, Furniss, & Walter, 2001), Williams Syndrome (Rodgers, Riby, Janes, Connolly, & McConachie, 2012a) and non-autism individuals with learning disabilities (Gadow et al., 2004; Gillot & Standen, 2007). Two studies have compared anxiety levels between autism and clinically anxious individuals (Russell & Sofronoff, 2005; Farugia & Hudson, 2006) and in general, found that the two groups showed comparable levels of anxiety. However, Russell and Sofronoff (2005), using the Spence Anxiety Scales (both parent and children version), found that while children with AS (aged years) and clinically anxious comparison group had similar anxiety scores on childrens report, parents of these children reported that ASD children had higher total anxiety scores as well as higher obsessive-compulsive and physical injury fears scores. This study points to the issue of discrepancy between parents and children s reports of anxiety that will be further discussed in the measurement section. In addition to comparing the levels of anxiety between autism group and TD controls and/or other clinical groups, several studies have looked at differences between levels of anxiety across different subtypes of autism (Muris, Steerneman, Merckelbach, Holdrinet, & Meesters, 1998; Tonge, Brereton, Gray, & Einfeld, 1999; Kanai, Koyama, Kato, Miyamoto, Osada, & Kurita, 2004; Weisbrot et al., 2005; Pearson et al., 2006; Thede & Coolidge, 2007; Kuusikko et al., 2008; Davis et al., 2011; Matson, Hess, & Boitsjoli, 2010) with mixed findings. In a large study that looked at anxiety in 182 preschool and 301 school children with AS, AD and PDD-NOS, Weisbrot et al. (2005) found no significant differences between 3 autism subgroups, either on parent or child version of Spence anxiety scale. However, in an older sample (6-12 years), children with AS showed significantly higher levels of anxiety than both PDD-NOS and AD children and PDD-NOS children had higher levels when compared with AD children. Three other studies compared levels of anxiety between AS and HFA individuals (Tonge et al. 1999; Thede & Coolidge, 2007; Kuusikko et al., 2008). A study by Tonge et al. (2008) found that AS group had higher levels of anxiety than HFA group. These results were partially supported by the findings of Thede and Coolidge (2007) whose results showed that while AS and HFA individuals had similar levels of overall anxiety the AS group had significantly higher scores for generalized anxiety. However, Kuusikko et al. (2008) found no differences in levels of anxiety between AS and HFA individuals. Four other studies compared PDD-NOS and AD group, with three studies finding higher levels of anxiety in PDD-NOS group (Muris et al., 1998; Davis et al., 2010; Matson et

44 44 al., 2010) and one study finding no differences (Pearson et al., 2006). Finally, Kanai et al. (2004) compared 53 HFA children with HFA (mean age= 6 years) and 21 children with AD (mean age: 8.2 years) using CARS-TV questionnaire and reported that individuals with HFA had significantly higher levels of anxiety. From the brief overview of the literature provided above it is clear that there is an agreement between the studies regarding the fact that anxiety is more prevalent in individuals with autism than in both general population and various clinical groups (apart from the clinically anxious group). However, results of anxiety prevalence studies have been much less consistent and this has been complicated by trying to divide individuals with autism into diagnostic subgroups, using different measures and including different age groups. In one of the first studies that examined the presence of co-occurring anxiety symptoms in autism, Muris et al. (1998) reported that 84.1% of children met criteria for at least one DSM anxiety disorder. Muris and colleagues used the Anxiety Disorders section of the Diagnostic Interview Schedule for Children (DISC; Version 2.3; National Institute of Mental Health, 1992) and their sample included 44 children with autism aged between 2 and 18 years. In contrast, Kim et al. (2000), using the Ontario Child Health Study Questionnaire Revised, found that only 13.6% of 59 children with autism (aged 9-14 years) met the criteria for elevated anxiety. However, most of the studies found the prevalence of anxiety to be between 35 and 55% (Bellini, 2004; Bradely et al., 2004; Gadow et al., 2005; Leyfer et al., 2006; De Bruin, Ferdinand, Meester, De Nijs, & Verheji, 2007; Sukhodolsky et al., 2007; Kuusikko et al., 2008; Simonoff, Pickles, Charman, Chandler, Loucas, & Baird, 2008; Mattila et al., 2010; Rodgers, Glod, Connolly, & McConachie, 2012; Strang et al., 2012). To briefly summarize, studies looking at the prevalence of anxiety subtypes in ASD have been inconsistent but overall, it seems that the most frequent types of anxiety disorders are specific phobias, generalized anxiety disorder, separation anxiety disorder, obsessive compulsive disorder, and social phobia (Muris et al., 1998; Evans et al., 2005; Gadow et al., 2005; Weisbrot et al., 2005; De Bruin et al., 2006; Gillot et al., 2007; Sukhodolsky et al., 2007; Mazefsk, Kao, & Oswald, 2011) Moderators of anxiety It has been suggested that in typical population, types of anxiety symptoms change during different stages of development (Warren & Sroufe, 2004). In general, separation anxiety and phobias of animals are prevalent in children between 6 and 9 years of age,

45 45 generalized anxiety symptoms and death phobias in children aged years, and social anxiety in adolescents aged 14 to 17 years. Several studies have supported this model. For example, Kashani and Orvaschel (1990) found that symptoms related to social anxiety steadily increased between the ages of 8 and 17. Following these findings from the typically developing population several studies have examined the role of age, developmental level in the expression of anxiety in autism, with mixed findings. Several studies found evidence that higher cognitive ability was associated with higher levels of anxiety (Weisbrot et al., 2005; Lecavalier, 2006; Mazurek & Kanne, 2010). For example, Weisbrot et al. (2005) found that anxious children with autism (n= 44) had higher IQ scores than non-anxious children with autism (n= 62). However, other studies failed to find association between IQ levels and anxiety (Sukhodolsky et al., 2007; White et al., 2009; Mazefsky et al., 2011; Strang et al., 2012). The findings regarding the relationship between chronological age and anxiety in autism have also been inconsistent, with some studies finding no association between chronological age anxiety (Sukhodolsky et al., 2007; Strang et al., 2012) and other studies suggesting that anxiety severity increases with age (Kuusikko et al., 2008; Davis et al., 2010, Green et al., 2012). It is important to point out that these findins are limited by the fact that very few studies have investigated anxiety in very you children and older adults with Autism. Three recently published studies (Davis et al., 2011; Matson et al., 2010; Green, Ben-Sasson, Soto, & Cater, 2012) have demonstrated that infants with autismand toddlers show more severe anxiety related symptoms than TD controls. Furthermore, Green et al. (2012) assessed anxiety in toddlers with autism at two time points, when they were aged months and 12 months later and found significant increase in anxiety between two time points. A recent cross-sectional study by Davis et al. (2011) compared the levels of anxiety in 131 toddlers (aged months), children (aged 3 16 years), and adults (aged years) with autism and found that anxiety levels increased from toddlerhood to childhood, decreased from childhood to young adulthood, and again increased from young adulthood into older adulthood, was reported. Although the study by Davis and colleagues is necessarily limited by its cross-sectional design and the use of different anxiety measures for different age groups, it does provide an initial insight into how anxiety might change over the life-span in individuals with autism.

46 Relationship between core autism symptomatology and anxiety Very few studies have examined the relationship between the social and communication difficulties, and anxiety difficulties. This is surprising if we have in mind that research on anxiety in the general population has consistently shown positive association between the impairments in language and communication and levels of anxiety (Beitchman et al., 2001; Pinborough-Zimmerman, Satterfield, Miller, Bilder, Hossain, & McMahon, 2007; Bornstein, Hahn, & Suwalsky, 2013), and considering the fact that the problems in communication and reciprocal social interaction are core features of autism. A study by Sukhodolosky et al. (2007) showed that anxiety levels were correlated with all three core autism symptoms. More precisely, increased levels of anxiety were significantly associated with higher language levels and also significantly, but negatively associated with non-verbal communication. The relationship between repetitive behaviours and anxiety found in this study will be discussed in depth in section Sukhodolsky and colleagues, similarly like Tsai (1996) and Strang et al. (2012) suggested that children with very basic communicative abilities may be less able to communicate their fears and anxieties and that language and communication problems affect expressed, rather than experienced levels of anxiety. In other words, lower levels of anxiety found in children without or with very limited language/communication abilities might be a consequence of children not being able to verbalize their problems rather than a true reflection of lower levels of anxiety in this population. Studies examining the relationship between overall autism severity (defined as total ADI-R scores) and anxiety have yielded mixed results with Mazurek and Keane (2010) finding a significant negative association between autism severity and anxiety levels and Lopata et al. (2010) reporting no significant association. In summary, mixed findings regarding the relationship between anxiety and core autism symptoms do not yet lend strong support to those theories of anxiety that propose that problems in social functioning are the main contributing factor for the development of anxiety in this population. A brief overview of the current theories of anxiety in autism will be summarized in the section below Theories of Anxiety in autism Numerous authors have suggested that the problems in social functioning, a core feature of autism, contribute to the development of anxiety in this population (Wing, 1992; Attwood, 1998; Tantam, 2000; Bellini, 2004; 2006). It has been hypothesized that early

47 47 deficits in social functioning can limit the individual s opportunity to interact with and learn from others, interfering with the ability to establish friendships and romantic relationships. Repeated social failures would cause increased emotional pain, leading to increased isolation which would then contribute to the development of anxiety (Wing, 1992). This model does provide plausible explanation for the presence of anxiety in individuals with autism who are cognitively able enough to realize their own problems and failures in the area of social interaction. However, as noted above, anxiety has also been found in younger and less cognitively able individuals with autism who are not likely to be aware of their own social deficits or be bothered by them. Anxiety in this population has been suggested to be a consequence of the constant overreactions to insignificant environmental sensory stimuli (Green & Ben-Sasson, 2010; Pfreiffer, Kinnealey, Reed, & Herzberg, 2005). It is likely the combination of both social deficits and sensory problems together with other cognitive deficits that puts individuals with autism at the risk for the development of anxiety. Because of problems in understanding of emotional states and intentions of others, together with over-reaction and inability to block-out insignificant environmental sensory stimuli, individuals with autism might experience their world as highly unpredictable, making even the most simple situations and events uncertain and difficult. Support for this hypothesis comes from the recent exploratory study by Ozsivadjian, Knott & Magiatti (2012). Ozsivadjian et al. (2012) fround that all five groups of parents of children and adolescents with autism provided fairly similar descriptions of the triggers for anxiety. Ozsivadjian and collegaues classified these triggers into the following categories: change or routines; social or language-related triggers; specific fears or phobias; sensory triggers; triggers relating to obsessions; and triggers relating to high expectations in performance or organization. In reflecting on findings from the literature discussed so far, my proposal would be that what most of these triggers have in common is the element of change and unpredictability. This indeed makes sense in the light of the general anxiety literature that identifies negative reactions when dealing with uncertain and unpredictable situations as one of the key factors in the development of anxiety (McEvoy & Mahoney, 2011). However, it is important to note that theories briefly summarized here have not been extensively tested and when testing has been attempted, results have been inconsistent. It can be argued that, similarly like with repetitive behaviours, our understanding of anxiety in autism is hindered by the lack of appropriate measures. The section below will very briefly

48 48 identify the key issues in the way that anxiety was assessed in the autism studies conducted thus far. These issues will be discussed in more depth in Chapter Main issues in the measurement of anxiety in population with autism Several authors have suggested that one of the key reasons for the conflicting findings in the literature on anxiety in autism, is the fact that at present, there is a lack of appropriate assessment of anxiety in this population (White et al., 2009; MacNeill, Lopes, & Minnes, 2009; Reaven, 2010; van Steensel et al., 2011). At present, the following six questionnaires and interviews have been designed to assess a broad range of pathologies (including, but not limited to anxiety) in individuals with developmental disabilities: Diagnostic Assessment for the Severely Handicapped-II (DASH-II; Matson, 1995), Autism Comorbidity Interview Present and Lifetime Version (ACI-PL; Leyfer et al., 2006), Autism Spectrum Disorders Comorbid for Children (Autism-CC; Matson & Gonzalez, 2007), Baby and Infant Scale for Children with Autistic Traits (BISCUIT; Matson, Boisjoli, & Wilkins, 2007), Behavioural Assessment System for Children-2 (BASC-2; Reynolds & Kamphaus, 2004). These assessment will be reviewed in more detail in Chapter 4 but in general, although some of these measures provide a fairly comprehensive assessment of anxiety related symptoms, their use is limited to either very young children or individuals with a low cognitive level and therefore not appropriate for older and more high functioning individuals with autism. Currently, there are no measures designed specifically for the assessment of anxiety in autism population. However, it is worth emphasizing the point made by Wood and Gadow (2010) that anxiety in autism should not be considered in isolation, in other words, it is necessary to compare phenomenology of anxiety in autism and general population. In order to do this, it is debatable whether, at this point, we necessarily need a measure developed specifically for autism or is it a more fruitful approach to carefully select a measure that is currently considered as a gold standard in the anxiety assessment in other populations and validate it in population with autism. The use of the same (or slightly modified) assessment across different populations will enable us to first identify common features of anxiety in autism and other populations and then to apply existing knowledge on anxiety disorders to this problem in autism. It is surprising thatthe majority of studies that examined various aspects of anxiety in autism did not employ measures that are currently considered to be a gold standard in the anxiety assessment in the general population. In fact, most of the studies have used assessments whose psychometric properties remain unexplored in general

49 49 population, let alone in autism population. Also, studies that used measures standardized in the general population failed to examine or at least to report their psychometric properties when used with population with autism. The second issue related to the assessment of anxiety in autism is on whose report (child s or caregiver s) to rely on. Individuals with autism are presented with an unique set of difficulties such as impairments in communication, problems with emotional self-awareness (Baron-Cohen et al., 2002; Losh et al., 2006) that might limit their ability to first identify and then articulate anxiety-related issues that they may face. However, this issue remains largely unaddressed in autism anxiety research Summary Research conducted so far confirmed the initial observations by Leo Kanner and Hans Asperger about the association between autism and anxiety by demonstrating that anxiety is a very frequent problem in this population. Results of studies looing at the relationsip between chronological and developmental age and anxiety levels in population with autism have not been consistent so far. We still do not know whether communication and social problems in autism play a crucial role in the development of anxiety in this population or, as suggested by Tsai (1996), have no or little influence on the actual levels of anxiety but only on the ability of individuals with autism to report anxiety. It is clear that some of the inconsistencies in the research on anxiety in autism are due to the problems of how to best assess anxiety in this population. Two major issues related to the assessment of anxiety are: the choice of appropriate measure and the informant issue. Although the need to design autism specific anxiety measures has been stressed by several authors (White et al., 2009; van Steensel et al., 2012), as discussed above, this thesis will argue that before designing such measure, it is first necessary to use anxiety measures that are considered as a gold standard for the anxiety assessment in general population and examine their psychometric properties in population with autism before opting for an autism specific measure. The issues of anxiety measurement, anxiety prevalence and the relationship between anxiety and chronological age, social and communication difficulties will be explored in Chapter 4. The relationship between anxiety and sensory problems will be addressed in Chapter Sensory Problems in Autism Initial description of what would currently be considered sensory symptoms date back to Leo Kanner (1943). Fascination with various sensory stimuli such as light reflecting from various surfaces as well as increased sensitivity and extreme

50 50 distress due to various sounds, moving objects or covering one s face are just some of the various atypical reactions to sensory stimuli that Kanner observed in 7 of his 11 original cases. Although Kanner himself (Kanner, 1943) rejected sensory hypothesis of autism (Rogers & Ozonoff, 2005), several pioneers in autism research did consider the possibility that sensory problems are behind some of the core autism features (Bergman & Escalona, 1949; Hut, Hutt, Lee & Ounsted, 1964; Ornitz, 1974; Ornitz, Guthrie, & Farley, 1977; Zentall & Zentall, 1983; Lovaas, Newsom, & Hickman, 1987). Kanner s early descriptions were corroborated by the subsequent clinical observations, autobiographical accounts and research studies. For example, in one of the earliest studies that systematically looked at the sensory processing in children with autism, Lorna Wing (1969) found responses of children with autism to sound, visual, tactile and olfactory stimuli to be markedly different from reactions of typically developing children, children with Down syndrome, receptive and expressive aphasia. Wing noted that responses of children with autism to different sensory stimuli resembled those of deaf-blind children. Strong, unpleasant reactions to sounds, visual and tactile stimulation and other sensory abnormalities are a prominent and recurring theme running throughout autobiographical accounts of individuals with autism as described here (Williams, 1994; O Neil, 1999; Grandin, 1992; Jones, Quigney, & Huws, 2003). The sharp sounds and bright lights were more than enough to overload my senses. My head would feel tight, my stomach would chum and my pulse would run my heart ragged until I found a safety zone. (Willey) I was also very touch sensitive; scratchy petticoats felt like sandpaper ripping off my skin. There is no way a child is going to function in a classroom if his or her underwear feels like it is full of sandpaper. (Temple Grandin) The sensory overload caused by bright lights, fluorescent lights, colours and patterns makes the body feel like being attacked or bombarded. (Williams) It is obvious that these problems present significant obstacles in the lives of people with autism. Despite this, sensory issues are currently not part of the diagnostic criteria for ASD (APA, 2000; WHO, 1944) but they are included in the latest version of the DSM-V criteria

51 51 (5 th ed.; DSM-V; APA, 2013). Inclusion of sensory problems in the latest diagnostic criteria are not at all surprising. As pointed out by Wing, Gould and Gillberg (2011) enough evidence accumulated over the years to support such change. Firstly, the majority of studies have demonstrated that more than 90% of autistic individuals exhibit various sensory problems (Rogers & Ozonoff, 2005; Leekam, Nieto, Libby, Wing, & Gould, 2007; Ben-Sasson, Hen, Fluss, Cermak, Engel-Yeger, & Gal, 2009). Secondly, several studies have found that atypical reactions to sensory stimulation in children as young as 12 months predicted later diagnosis of autism (Osterling & Dawson, 1994; Baranek et al., 1999). For example, Baranek and colleagues (Baranek et al., 1999) found that decreased orienting to visual stimuli, increased mouthing of objects and aversion to touch distinguished 9-12 months infants who were later diagnosed with autism from the ones who did not receive diagnosis. Finally, sensory problems have been found to be stable across the life-span of individuals with autism. In a follow-up study with 105 individuals with autism and adults who were diagnosed as children, Billstedt, Gillberg, & Gillberg (2007) found that all individuals showed sensory problems during original diagnostic assessment and more importantly, that 93% of those individuals continued to have impairing sensory abnormalities as adults. Billstedt and colleagues used the Diagnostic Interview for Social and Communication disorders (DISCO) that provides detailed information on sensory problems during both assessments. Despite the significant progress in the understanding of sensory problems in autism it is still not clear which sensory problems are specific to autism, how sensory problems are associated with the core autism symptomatology and how sensory problems change with age. This section will review the literature relevant to these topics. Before this, it is important to highlight that the term atypical sensory processing has been used in inconsistent ways throughout the literature. Some authors referred to what is defined as Sensory Discrimination Disorder by the Interdisciplinary Council on Developmental and Learning Disorders (ICDL; 2005; see also Miller, Anzalone, Lane, Cermak, & Osten, 2007), in other words, to the difficulties in interpreting qualities of sensory stimuli and/or inability to perceive similarities and differences among stimuli (ICDL, 2005). Other authors have used the term atypical sensory processing to denote the difficulties in responding to sensory input with behaviour that is appropriate to the nature and/or intensity of the sensory stimuli. These difficulties are termed as Sensory Modulation Disorders (ICDL, 2005) and throughout this thesis, the term atypical sensory processing will refer to the Sensory Modulation Disorder. Although Sensory Discrimination Disorder will not be the topic of this thesis, for the sake of

52 52 comprehensiveness, a short overview of these difficulties will be provided before moving to the issues of prevalence, specificity and phenomenology of Sensory Modulation Disorders in autism Sensory Discrimination Disorder Atypical sensory discrimination is well documented across all modalities in autism. Behavioural and electrophysiological studies showed both enhanced and impaired performance in different areas of visual processing. Higher levels of visual acuity in autism group when compared to TD participants were found by Ashwin, Ashwin, Rhydderch, Howels, & Baron-Cohen (2009). However, several other studies failed to replicate these findings (Bolte et al., 2012; Tavassoli, Latham, Bach, Dakin, & Baron-Cohen, 2011). Results regarding colour discrimination, biological motion perception and visual search seem to be more consistent with several studies reporting impaired colour discrimination in individuals with autism (Franklin, Sowden, Burley, Notman, & Adler, 2008; Heaton, Ludlow, & Roberson, 2008), intact biological motion perception apart from processing emotional expressions (Atkinson, 2009; Murphy, Brady, Fitzgerald, & Troje, 2009; Kaiser, Demolino, Tanaka, & Shiffrar, 2010; Koldewyn, Whitney, & Rivera, 2010) and superior visual search performance (O Riordan, Plaisted, Driver, & Baron-Cohen, 2001; Jarrold, Gilchrist, & Bender, 2005). Atypical responses to auditory stimuli are frequently reported in questionnaire studies (Kientz & Dunn, 1997; Talay-Ongan & Wood, 1999; Dunn, Myles, & Orr, 2002; Rogers, Hepburn, & Wehner, 2003; Tomchek & Dunn, 2007, Baker, Lane, Angley, & Young, 2008; Wiggins, Robins, Bakeman, & Adamson, 2009; Hilton, Harper, Kueker, Lang, Abbacchi, Todorov, & LaVesser, 2010, Klintwall et al., 2010). Behavioural studies have consistently found increased pitch perception for both simple and complex tones as well as speech sounds (Bonnel, Mottron, Peretz, Trudel, & Gallun, 2003; Gomot, Belmonte, Bullmore, Bernard, & Baron-Cohen, 2008; Jones et al., 2009). Results regarding intensity discrimination seem less consistent with some studies finding evidence for hyper-acusis (Khalfa et al., 2004) and others finding no differences in intensity discrimination between the autism and TD groups (Jones et al., 2009). Lastly, evidence suggest that Individuals with autism are better at local processing of sounds (Mottron, Peretz, & Menard, 2000; Jarvinen-Pasley & Heaton, 2007). Several experimental studies have shown that Individuals with autism have higher thresholds for higher (but not lower) frequency and difference placement of tactile stimuli

53 53 (Blakemore, Tavassoli, Calo, Thomas, Catmur, Frith, & Haggard, 2006; Cascio et al., 2008) as well as lower pain thresholds for thermal tactile stimuli (Cascio McGlone, Folger, Tannan, Baranek, Pelphrey, & Essick, 2008). Systematic research on the olfactory and taste processing in autism has started only recently. Conflicting results have been found regarding the olfactory detection thresholds with Tavassoli and Baron-Cohen (2012a) finding no differences between autism and TD controls and Dudova, Vodicka, Havlovicova, Sedlacek, Urbanek, and Hrdlicka (2011), reporting impaired olfactory detection thresholds in group with autism. Both Tavassoli and Baron-Cohen (2012a) and Dudova et al. (2011) used the standardized Sniffin Sticks olfaction task. Mixed results have also been found regarding the odour identification (Suzuki, Critchely, Rowe, Howlin, Murphy, 2003; Dudova et al., 2011; May, Brewer, Rinehart, Enticott, Brereton, & Tonge, 2011). Only two studies have examined taste identification in autism finding that Individuals with autism were impaired at taste identification (Bennetto, Kuschner, & Hyman, 2007; Tavassoli & Baron-Cohen 2012b). Sensitivity to and fascination with smells have been constantly reported in questionnaire studies (Kientz & Dunn, 1997; Dunn et al., 2002; Rogers et al., 2003; Leekam et al., 2007; Tomchek & Dunn, 2007; Schoen, Miller, Brett-Green, & Nielsen, 2009; Wiggins et al., 2009; Hilton et al., 2010; Kintwall et al., 2010) Sensory Modulation Problems in autism Sensory modulation disorder (SMD) is defined as the failure to behaviourally respond in a way that is appropriate to the intensity and nature of sensory stimulation. Dunn (1997) proposed four SMD subtypes patterns: a. sensory hyper-sensitivity-characterized by distress to innocuous stimuli and passive withdrawal from unpleasant situations; b. sensation avoiding-also characterized by distress but also active avoidance; c. low registration (or sensory hypo-responsiveness)-characterized by either complete lack of or delayed response to a stimulus or a higher response threshold (e.g., only responding to a more intense stimuli); d. sensation seeking-behaviours that either perpetuate or intensify a sensory experience. A very similar classification of SMD has been proposed by the Interdisciplinary Council on Developmental and Learning Disorders (ICDL Work Groups, 2005). ICDL

54 54 proposed that SMD can be classified into: (a) Sensory Overresponsivity; (b) Sensory Underresponsivity; and (c) Sensory Seeking. As can be seen, the only difference between the two classification systems is that whereas Dunn s model distinguishes between sensory sensitivity and avoidance as two separate (though related) expressions of sensory overresponsiveness (SOR), ICDL classification considers SOR unitary construct Prevalence of Sensory Modulation Disorders in Autism Over 20 questionnaire studies have compared sensory responses of individuals with autism to both typically developed individuals and individuals with various disorders. In general, studies have consistently shown that Individuals with autism have more SMD problems when compared to TD individuals. Most of the studies used the Sensory Profile questionnaire. The Sensory Profile (SP; Dunn, 1999) is a 125 item, norm-referenced, parentreport questionnaire which measures a child s/young person s responses to everyday sensory experiences. Parent rate frequency of each items on a 5-point Likert scale ranging from always to never. SP assesses sensory processing across different modalities and can also be used to obtain four SMD patterns (or sensory quadrants in Dunn s terminology). In one of the earliest questionnaire studies, Kientz and Dunn (1997) used the original Sensory Profile to examine sensory processing in 38 children with autism (agerange: 3-13 years) and 64 TD children aged 3-10 yeas. It was found that 85% of the Sensory Profile items differentiated children with autism from TD children. Children with autism had significantly lower scores (indicating poorer performance) across all six modalities and on both sensory hyper and hypo-sensitivity. Consequent studies that used the Sensory Profile largely replicated results found by Kientz & Dunn in individuals with autism across all ages (Watling, Deitz, & White, 2001; Dunn et al., 2002; Smith Myles et al., 2004; Kern et al., 2007; Ben-Sasson, Cermak, Orsmond, Carter, Kadlec, & Dunn, 2007; Tomchek & Dunn, 2007; Crane, Goddard, & Pring, 2009; Hilton et al., 2010). For example, Crane et al. (2009) found that a group of 18 adults with autism (age range: years) had poorer performance in terms of overall Sensory Profile scores as well as on the hypo-sensitivity, sensory sensitivity and sensation avoidance quadrants, but not on the sensation seeking quadrant when compared to 18 age and IQ matched TD controls. In one of the rare studies that compared sensory responses of autism and TD toddlers, Ben-Sasson et al. (2007) found that autism toddlers (age range: months) had higher frequency of sensory under-responsiveness and avoidance, and lower frequency of sensory seeking behaviours than CA and MA matched TD controls. Differences

55 55 between autism and TD individuals in sensory processing found using Sensory Profile were largely replicated in studies that used other questionnaires/interviews, most notably Diagnostic Interview for Social and Communication Disorders (DISCO; Wing et al., 2002), Sensory Sensitivity Questionnaire (Talay-Ongan & Wood, 2000) and Sensory Experiences Questionnaire (Baranek, David, Poe, Stone, & Watson, 2006). For example, Leekam et al. (2007), using the sensory items from DISCO, found that more than 90% of children with autism had sensory abnormalities in at least one sensory domain. Relatively few studies have compared the sensory processing of individuals with autism with other non-typical populations. In a study that compared children with autism and ADHD, Ermer and Dunn (1998) found that ADHD children had higher incidence of sensory seeking and inattention/distractibility than children with autism who in turn had more hypersensitive reactions across different modalities and also scored higher on the fine motor/perceptual difficulties factor that is indicative of hypo-sensitivity. Rogers et al. (2003) using the Short Sensory Profile found that while children with autism had more sensory symptoms (especially increased sensitivity to tactile and taste/smell stimuli) than both DD and TD groups, they were not significantly different when compared to the children with Fragile X syndrome. Several other studies compared sensory processing of Individuals with autism and DD individuals and in general, individuals with autism were found to have more sensory problems than DD individuals (Baranek, David, Poe, Stone, & Watson, 2006; Leekam et al., 2007; Wiggins et al., 2009; Watson, Patten, Baranek, Poe, Boyd, Freuler, & Lorenzi, 2011). A very important question when comparing sensory problems in autism with both TD and various clinical groups is whether there is a particular type of sensory problems that is specific to the group with autism. Discussion below will consider findings regarding the specificity of sensory hyper-, hypo-sensitivity and sensory seeking separately Sensory hyper-sensitivity Evidence regarding the specificity of sensory hyper-sensitivity in individuals with autism have been somewhat inconsistent but in general, results seem to suggest that hypersensitivity is not specific to the population with autism. Several studies reported a higher frequency of sensory hyper-sensitivity in children with autism when compared to TD children (Baranek et al., 2006; Dunn, Smith-Myles, & Orr, 2002), however, other studies failed to find evidence that sensory hyper-sensitivity distinguished children with autism from either TD children (Ermer & Dunn, 1998; Watling et al., 2001), from children with various

56 56 developmental delays (Baranek et al., 2006) or individuals with ADHD (Ermer & Dunn, 1998). For example, using the Sensory Experiences Questionnaire, Baranek et al. found that although 56% of children with autism aged 5-80 months showed extreme sensory hypersensitive behaviours, these behaviours did not distinguish them from CA matched DD children. It has been suggested by several authors (e. g. Rogers & Ozonoff, 2005) that sensory hyper-sensitivity is a consequence of lower developmental level rather than specific to autism. However, it is important to point out that hyper-sensitivity is found widely outside of learning disability populations, for example in individuals with anxiety (Hofmann & Bitran, 2007; Farrow & Coulthard, 2012), Schizophrenia (Brown, Cromwell, Filion, Dunn, & Tollefson, 2002) and as pointed out above, ADHD (Ermer & Dunn, 1998) Sensory hypo-sensitivity Sensory hypo-sensitivity has been suggested to be distinctive of children with autism (Baranek et al., 2006; Rogers & Ozonoff, 2005). Findings from several studies that showed that individuals with autism exhibit a significantly higher frequency of hypo-sensitivity behaviours when compared to both CA matched TD individuals (Baranek et al., 2006; Dunn et al., 2002; Watling et al., 2001) and either TD or other clinical groups matched on mental age (MA) (Baranek et al., 2006) seem to lend support to the claim that sensory hyposensitivity is specific to autism. For example, Baranek, et al., (2006) found that 63% of children with autism showed sensory hypo-sensitivity and that a pattern that differentiated autism from both TD and DD groups was hypo-sensitivity to both social and non-social stimuli. However, several studies failed to find a support for hypo-sensitivity as distinguishing factor between individuals with autism and CA matched TD children, children with ADHD or children with Fragile X syndrome (Ermer & Dunn, 1998; Rogers et al., 2003). Furthermore, sensory hypo-sensitivity has also been found in clinical disorders without the co-morbid learning disability, for example in Schizophrenia (Brown et al., 2002) Sensory Seeking Results regarding sensory seeking behaviours seem to be the least consistent. While two studies failed to find the difference in sensory seeking behaviours between autism and TD groups, in both toddlers (Dunn, 2002) and adults (Crane et al., 2009), other studies found increased frequency of sensory seeking in children with autism aged 3-6 years (Watling et al., 2001) and in individuals with autism aged 3-43 years (Kern et al., 2006). All these studies used Sensory Profile. Studies that compared the group with autism with other clinical groups

57 57 also reported inconsistent results. For example, Ermer & Dunn (1998) found that the group with autism showed significantly lower frequency of sensory seeking behaviours than the children with ADHD. On the other hand, a study by Watson et al. (2011) found that toddlers with autism showed significantly more sensation seeking behaviours than CA matched DD children. As can be seen from the brief review of the literature, it is difficult to draw conclusions regarding the question of which, if any, type of sensory problems differentiates individuals with autism from both general population and other clinical groups. Hyperresponsiveness does not seem to be specific to autism as it can be seen in various other disorders and as suggested by Rogers and Ozonoff (2005) it may well be a consequence of lower developmental level rather than any specific pathology. However, as pointed out, sensory hyper-sensitivity can be found in several disorders that are not characterized by low developmental level, for example anxiety. It also seems that hypo-responsiveness might be more prevalent in the group with autism although, as reviewed above, the results are inconsistent. It is not possible to draw any conclusion regarding the specificity of sensory seeking behaviours in autism. Previous discussion regarding the specificity of particular type of sensory problems in the population with autism usually does not take into the consideration the fact that several studies have found that individuals with autism tended to show a mixed type of sensory responsiveness, in other words an individual would show atypical scores across all Sensory Profile quadrants i. e. sensory hyper-, hypo-sensitivity, seeking and avoidance (Baranek et al., 2006; Ben-Sasson et al., 2007; Tomchek & Dunn, 2007). It is possible that, like repetitive behaviours, it is the frequency of sensory modulation disorders rather than their systematic form that distinguishes individuals with autism from both TD individuals and other clinical groups. The simultaneous presence of both sensory hyper- and hypo-responsivness in autism has not been properly looked into so far. One line of research seems to suggest that a subgroup of individuals with autism shows hyper-responsiveness to environmental stimuli from infancy and toddlerhood. This state of permanent heightened responsiveness would lead to chronic stress with alteration in functioning of hypothalamic-pituitary-adrenal axis and its hormones, change in cytokines metabolism and change in the autonomic nervous system functioning. These changes, although generally adaptive in the short term, if persist over a longer period of time lead to neurotoxic effects to limbic system and permanent alterations in

58 58 endocrine system functioning. This model is supported by findings from the research on the relationship between anxiety and depression where it has been proposed that anxiety and a state of sensory hypersensitivity over time may precede a state of depression and hyposensitive responses to stimuli in the environment (Markram & Markram, 2010; McEwen, 1998; McEwen & Wingfield, 2010; Ganzel, & Morris, 2011). Other researchers have suggested more circular relationship between sensory hyperand hypo-sensitivity (Lane, Young, Baker, & Angley, 2010; Lane, Dennis, & Geraghty, 2011). In other words, it has been hypothesized that individuals with autism have problems in finding and maintaining levels of arousal in the middle ground between hyper- and hyposensitivity (Pfeiffer et al., 2005). Furthermore, it has also been suggested that sensory seeking and sensory avoidance behaviours might represent an attempt to modulate the current state of arousal, in other words, sensory seeking might represent a soothing mechanism when an individual is in the state of over-arousal, and for raising stimulation in cases of underarousal; and sensory avoidance might represent a mechanism for coping with the state of over-arousal and consequent over-stimulation (Lane et al., 2011). The two accounts described above will be explored in Chapter 5. At this point it is important to highlight the fact that although studies were very consistent in terms of the measure used-most studies using either the full or abbreviated Sensory Profile, there was a wide variability between the studies in terms of the characteristics of the group with autism with some studies being limited to older, higher functioning individuals and other studies including individuals with a wide age range and abilities. This, together with the fact that in some cases the group with autisms were matched with control groups on chronological age, and in other on developmental age necessarily limits generalizability of the findings. Indeed, it has been suggested that sensory problems can change over time (Rogers & Ozonoff, 2005; Ben-Sasson et al., 2009) and when considering sensory problems, it is necessary to take into account chronological age and developmental level. Before defining the questions that this thesis will try to answer, in the following section I will provide a brief review of the literature examining the association between sensory problems and chronological and developmental age as well as a small body of literature that looked at the relationship between core autism features and sensory problems.

59 The relationship between sensory modulation problems, chronological age and developmental level The results of the studies that examined occurrence of sensory problems across development have been very inconsistent. A cross-sectional study by Kern et al. (2006) reported that sensory problems were less frequent at later ages in their sample of individuals with autism (age range 3-56 years), that older individuals with autism became more similar to TD controls in the way that they respond to sensory stimuli. The only exception was tactile hyper-sensitivity that did not improve over time. However, other studies did not find evidence for the improvement in sensory problems over time (Talay-Ongan & Wood, 2000; Leekam et al., 2007). Using the Sensory Sensitivity Questionnaire with a group of 30 children with autism age between 4 and 14 years, Talay-Ongan and Wood (2000) found that sensory problems were more common in a subgroup of 10 to 14 year olds than in a subgroup of 6 to 9 year olds who in turn had more problems than 4 to 5 year olds. Leekam et al. (2007) found the evidence that, with the exception of visual symptoms, other sensory problems persisted over time supporting numerous autobiographical accounts which pointed out that sensory problems were an on-going problem regardless of the age of individual (Jones et al., 2003). Results from a relatively recent meta-analysis by Ben-Sasson et al. (2009) which summarized results from 14 studies suggest a non-linear course of sensory problems in autism with an increase in total sensory problems, sensory hyper-sensitivity and sensory seeking up to ages 6-9 and a decrease after the age of 9. Meta-analysis failed to find evidence for a consistent course of sensory hypo-sensitive behaviours. As has been pointed out in the section on the development of repetitive behaviours, although in typical population developmental age parallels chronological age, this is not necessarily the case in autism, and differences in the developmental level might have contributed to the mixed findings regarding the relationship between chronological age and sensory problems. However, results regarding the influence of cognitive level on the expression of sensory problems have also been very inconsistent. Rogers et al. (2003) failed to find an association between developmental level and any of the Sensory Profile scores. Similar findings were reported by several other studies (Kientz & Dunn, 1997; Baranek et al., 2006; Leekam et al., 2007; Baker et al., 2008; Lane et al., 2010). On the other hand, Crane et al. (2009) found that IQ level was negatively associated with three of four sensory profile quadrants-sensory hypo- and hyper-sensitivity and avoidance but not with sensory seeking behaviours.

60 Sensory Problems and Core Autism Symptoms The majority of research on sensory problems in individuals with autism has focused on comparing the frequency/severity of sensory probems between individuals with autism with both TD individuals and DD individuals and relatively few studies have examined the relationship between sensory problems and core autism symptoms. Several studies have reported associations between overall autism severity and the presence of more sensory problems (Adamson, O Hare, & Graham, 2006; Kern et al., 2006; Ben-Sasson, Cermak, Orsmond, Tager-Flusberg, Carter, & Kadlec, 2008). For example, using the CARS (Schopler et al. 1994) as a measure of autism severity, Kern et al. (2006) found that total Sensory Profile scores correlated with autism severity in children with autism. It is interesting to note that Kern and colleagues did not find correlations between sensory problems and autism severity in adolescents and adults. However, from looking at the overall autism severity, it is difficult to interpret the relative contribution of individual core symptoms to the significant correlations found in the previously mentioned studies. A detailed review on the relationship between repetitive behaviours and sensory problems will be provided in Section of this chapter. In this section I will focus on the studies that have examined the relationship between sensory problems and other two core autism features. Only four studies have examined the association between social/communicative deficits and sensory problems (Liss et al., 2006; Hilton, Graver, & LaVesser, 2007; Hilton et al., 2010; Watson et al., 2011). Hilton et al. (2007) compared the levels of sensory problems assessed by the Sensory profile between a group of individuals with autism with Social Responsiveness Scale (SRS; Constantino & Gruber, 2005) score in the severe range of functioning and a group of individuals with SRS score that suggested milder impairments. It was found that thw individuals with autism with higher SRS scores had significantly more problems on all 4 Sensory Profile quadrants and that sensory hyper-sensitivity and avoidance had strong negative correlations with SRS total scores while sensation seeking and hypo-responsiveness were moderately negatively correlated. Similar findings were reported by Watson et al. (2011), apart from the fact that they did not find significant correlations between social/communication problems and sensory hyper-sensitivity. Finally, Liss, Saulnier, Fein, & Kinsbourne (2006) found positive correlations between both hypo-responsiveness and sensory seeking, and social and communication symptom severity in children and adults with autism spectrum disorders. Hyper-responsiveness did not correlate with either social or communication symptoms.

61 Neurobiology of Sensory Problems in Autism Despite the prevalence and clinical importance of sensory problems in autism, there is a relative paucity of studies exploring the neurobiological basis of these problems. In one of the few functional magnetic resonance imaging (fmri) studies on the processing of nonsocial sensory stimuli in autism, Gomot, Belmonte, Bullmore, Bernard and Baron-Cohen (2008) found that children with autism (N= 12, mean age= 13.5 years, SD= 1.6) had higher activity in the right prefrontal-premotor and the left inferior parietal regions when presented with novel auditory stimuli than typically developing children (N= 12, age and IQ matched to ASD children). Interestingly, there were no group differences in the primary auditory cortex activity. On a behavioural level, children with autism were faster than TD children in their ability to detect novel auditory stimuli. Gomot and colleagues suggested that although there were no differences in terms of initial registration of novel auditory stimuli (at the level of primary auditory cortex), these stimuli were differently processed (attended) by higher prefrontal regions. fmri evidence of hyperactivity to auditory stimuli are also supported by electrophysiological findings (Gomot, Giard, Adrien, Barthelemy, & Bruneau, 2002; Gomot, Blanc, Clery, Roux, Barthelemy, & Bruneau, 2011; but see Ceponiene et al., 2003; Lepisto et al., 2006). For example, Gomot et al. (2011) used the mismatched negativity (MMN) and P3a event related potentials (ERP) to explore the neural basis of novelty auditory detection in 27 children with autism (age range: 5-11 years) and 27 gender, age and IQ matched TD children. It was found that ASD children showed significantly shorter MMN latency and greater P3a amplitude when compared to TD children. In a very recent study Green and colleagues (Green et al., 2013) used fmri to explore brain responses of 25 individuals with ASD (mean age= years, SD= 2.29) and 25 matched TD controls to mildly aversive sensory stimuli (presented in visual and auditory modality). It was found that ASD group showed increased responses in the primary visual and auditory cortex, thalamus, hippocampus, amygdala and orbitofrontal cortex suggesting that sensory problems might be the results of impairments in both bottom-up as well as in top-down processing. In addition to the neuroimaging methods reviewed above, another approach to exploring the biological basis of sensory problems is the use of non-invasive measures of activity of sympathetic and parasympathetic branches of autonomic nervous system. An influential paper by Rogers and Ozonoff (2005) provided a systematic assessment of the literature on physiological responses to sensory stimuli in individuals with ASD and despite inconsistencies in the literature (for example Bernal and Miller (1970) and van Engeland

62 62 (1984) reported reduced activity and Stevens and Gruzelier (1984) increased activity of sympathetic nervous system) concluded that ASD individuals are characterized by hypoactivity to sensory stimuli. It is important to emphasise that physiological studies reviewed by Rogers and Ozonoff were published before 2000 (and majority in 1970s and 80s) and were limited in terms of both the equipment used and by the fact that some of the studies were published prior to the DSM-III-R classification which raises some concerns about the reliability of the diagnosis. However, in recent years several studies have explored the profile of autonomic nervous system responses to sensory stimuli in ASD individuals and these studies will be summarized below. In a study that explored the activity of sympathetic and parasympathetic systems in 59 children with ASD (mean age= 92 months, SD= 13 months) and 29 typically developing children (mean age= 98 months; SD= 15 months), Schaaf, Benevides, Leiby and Sendecki (2013) used the Sensory Challenging Protocol to present sensory stimuli. The Sensory Challenging Protocol (McIntosh, Miller, & Shuy, 1999) consists of 3 minute baseline condition followed by the presentation of 48 sensory stimuli (8 stimuli in each of the primary sensory domains) with each stimulus lasting 3 seconds. The stimuli are divided with variable seconds inter-stimulus interval. Respiratory sinus arrhythmia (RSA), naturally occurring heart rate variability during respiration, was used as an index of parasympathetic system activity and pre-ejection period (PEP), the interval from the beginning of electrical stimulation of the ventricles to the opening of the aortic valve, was used as an index of sympathetic system activity. While TD participants exhibited a decrease in RSA in response to challenging stimuli. suggesting intact functioning of the vagal brake (Porges, 2007), this pattern was not evident in children with ASD. This suggests that children with ASD had problems in utilizing the activity of parasympathetic system in regulating their responses to sensory stimulation. Although there children with ASD did not differ from TD children in terms of their sympathetic activity, PEPs were lower in ASD children (with high effect sizes for auditory, olfactory and movement domains of the Sensory Challenging protocol) suggesting higher activity of the sympathetic system in ASD children. Woodard et al. (2012) compared the heart rate reactivity to sensory stimulation between 8 children with ASD and 8 TD children (both groups had the age range between 2 and 3.15 years). The sensory protocol consisted of the presentation of visual, auditory, olfactory, tactile and taste stimuli, 22 in total. Stimulus presentation was counterbalanced (with the exception of the taste stimuli which was always presented at the end of the protocol) with 30s rest interval between each

63 63 stimulus presentation. Heart rate responses were significantly higher in children with ASD during both baseline and across sensory stimuli presentations suggesting sympathetic stimuli hyper-activity. Unlike Schaaf et al. (2013) and Woodard et al. (2012) who found evidence of increased autonomic nervous system reactivity to sensory stimuli, Schoen, Miller, Brett- Green and Nielsen (2009) reported evidence for under-reactivity of ANS system. They found that skin conductance responses during the Sensory Challenging Protocol were lower in ASD children (N= 40; mean age= 9.3 years, SD= 2.74) when compared to both typically developing children (N= 33, mean age= 8.1 years, SD= 2.44) and children with Sensory Modulation Disorder (N= 31, mean age= 8 years, SD= 1.93). Inconsistent findings highlighted above can be explained by the results from the study by Schoen, Miller, Brett-Green and Hepburn (2008) that suggested the existence of two subgroups of ASD children (total N= 40, age range: 5-15 years) based on their pattern of skin conductance responses during the sensory stimulation. One group of children was characterized by low skin conductance levels amplitude and variability (slower latency and faster habituation) and the other with high skin conductance levels amplitude and more SCR variability (faster latencies and slower habituation). These findings were in line with previous findings from Hirstein, Iversen and Rachandran (2001) who identified that around 30% of their ASD participants showed low baseline skin conductance and low reactivity to stimulation while the rest of their sample had high baseline arousal levels and were hyperactive to sensory stimuli. In summary, despite the strong evidence (reviewed in previous subsections) suggesting that majority of ASD individuals exhibit some form of sensory problems, neurophysiological basis of these problems are currently not clear. What is clear from the brief overview of the literature is that future work will need to develop more carefully designed, ecologically valid experimental sensory protocols. Furthermore, more dimensional approach to ASD combined with exploration of neurophysiological basis of sensory problems across other conditions will be necessary in order to gain better understanding of biological basis of sensory problems. Some of the future directions will be discussed in chapter 10 of this thesis.

64 Summary As can be seen from the short review provided above, one consistent finding from the literature on sensory problems in autism is that sensory atypicalities are a very frequent problem in this population. However, it is still not clear which of these problems is the most pervasive problem in autism, what is the relationship between sensory modulation problems and chronological age, developmental level, and core autism features. Also, previous research has not addressed the question of what is the mutual relationship between different sensory modulation problems in autism. Having all this in mind, Chapter 5 of this thesis will aim to provide a detailed description of sensory processing in a sample of children and adolescents with autism. It will also examine the mutual relationship between sensory hyper-, hyposensitivity, sensory avoidance and sensory seeking types of sensory responsiveness. The role of chronological age, language level and core autism symptomatology will be examined. Also, the influence of different sensory modulation difficulties on adaptive abilities will be examined Repetitive behaviours, sensory problems and anxiety in autism I have suggested that the bizarre and repetitive motor output may actually be a compensatory activity which helps the autistic child to make sense out of sensation. (Ornitz, 1974, p203). The potential role of repetitive behaviours as a way of controlling sensory unpredictability and warding off anxiety was hinted in the original descriptions of Leo Kanner (Kanner, 1943). He noted that the child s behaviour is governed by an anxiously obsessive desire for the maintenance of sameness...changes of a pattern, of the order in which every day acts are carried out, can drive him to despair...the dread of change and incompleteness seems to be a major factor in the explanation of the monotonous repetitiousness and the resulting limitations in the variety of spontaneous activity. More detailed account of the role of repetitive behaviours as a means for providing internallycontrolled mechanisms for maintaining a homeostatic state of arousal was developed by Hutt, Hutt, Lee, and Ounsted (1964). Hutt and colleagues suggested that over-arousal and hypersensitivity to sensory stimulation led to the development of stereotypical motor behaviours. This hypothesis was based on several lines of evidence, in particular a series of studies using

65 65 EEG which suggested stereotypical pattern of behaviours provided a means of avoiding sensory stimulation. A more comprehensive account of the relationship between sensory problems, arousal and repetitive behaviours was provided by Ornitz and Ritvo (1968) and later by Kinsbourne (1980). Ornitz and Ritvo suggested that the problems in sensory integration and maintenance of optimal levels of arousal were the primary symptoms of autism which led to the development of repetitive behaviours and problems in language, social and other abnormalities which were considered as secondary symptoms. Both Ornitz and Ritvo as well as Kinsbourne suggested that individuals with ASD, due to brainstem abnormalities, were constantly fluctuating between the states of over-excitation and over-inhibition and suggested that repetitive movements could be compensatory strategies for both of these states rather than compensatory strategies for just one or the other. These authors hypothesized that the individual with autism is likely to experience sensory problems as distressing and that due to this the individual may subconsciously search for ways to calm down and reduce the unpredictability by insisting on rigid patterns of routines. Such hypotheses about the role of repetitive behaviours as a coping mechanism for dealing with problems in regulating and maintaining optimal levels of arousal and consequent hyper/under-responsitivity and anxiety, have been supported by findings from the studies in typical population. For example, motor stereotypies such as leg swinging, rocking, repetitive finger movements and nail biting are often seen in typically developed individuals in the situations of boredom/understimulating environment with a function of increasing stimulation (Mason, 1991) and in demanding situations where they have been hypothesized to serve to block excessive stimulation (Wehmeyer, 1989). A potential link between stereotyped movements and sensory processing is further supported by the fact that these behaviours are observed in individuals with hearing and visual impairments (Troster, Brambring, & Beelmann, 1991, Gal, Dyck & Passmore, 2002; 2010). Troster et al. (1991) used the Bielefeld Parents' Questionnaire for Blind and Sighted Infants and Preschoolers to examine the frequency and duration of stereotypic behaviours in 85 blind children aged months and found that all children exhibited at least one type of motor stereotypy with body rocking and eye pocking being the most prevalent. They also found that heightened arousal and monotony were the typical situations when stereotypies occurred. It has been hypothesized that ritualistic and compulsive behaviours in early childhood have emotion regulation role and serve to ward of anxiety (Gesell et al., 1974; Evans et al.,

66 ). The link between childhood rituals and anxiety are very interesting in the light o the fact that some of these rituals are very similar to OCD symptoms which have also been hypothesized to be associated with anxiety Evans et al., 1997; Mataix-Cols, & Rosario- Campos, & Leckman, 2005). However, despite early theories suggesting the relationship between repetitive behaviours, sensory problems and anxiety in individuals with autism, and above mentioned findings from the typical development and other conditions where repetitive behaviours are a prominent or even a diagnostic feature, this relationship has not been thoroughly examined in the autism literature. Systematic review of the literature conducted for this thesis (with January 2013 as the last date of search) has identified 17 studies that examined the relationship between repetitive behaviours, sensory problems and anxiety. The relationship between repetitive behaviours and sensory problems was explored in 8 studies, between repetitive behaviours and anxiety in 5 studies and between anxiety and sensory problems also in 4 studies. At the time of writing this thesis, no published studies have examined the three way relationship between repetitive behaviours, sensory problems and anxiety. Studies identified in the literature search will be briefly reviewed below and described in more detail in Chapter Repetitive behaviours and sensory problems As stated above, 8 published studies have addressed the question of the potential relationship between repetitive behaviours and sensory problems. Three studies (Gabriels et al., 2008; Chen, Rodgers, & McConachie, 2008; Boyd, McBee, Holtzclaw, Baranek, & Bodfish, 2009) have explored the relationship between sensory problems and repetitive behaviours (RRBs were treated as a unitary category) and in general, found that higher levels of repetitive behaviours were associated with more sensory problems. For example, Gabriels et al. (2008) found a positive relationship between repetitive behaviours and sensory processing problems in a sample of 70 children with autism, aged years. This relationship stayed significant after controlling for IQ. Gabriels and colleagues also identified a subgroup of participants with autism that had high rates of both sensory processing problems, repetitive behaviours and high rates of co-morbid psychiatric problems, and did not differ from other participants in terms of chronological age and IQ. Despite providing support for the hypothesis that higher levels of total repetitive behaviours are associated with more sensory problems, the study by Gabriels et al. (together with studies by Boyd et al and

67 67 Chen et al. 2008) is limited in the sense that it did not examine whether different subtypes of repetitive behaviours are differently associated with different sensory behavioural patterns. This limitation was addressed by Baranek, Foster and Berkson (1997), Boyd et al. (2010), Gal et al. (2002) and Gal et al. (2010). However, results of these studies have been somewhat conflicting with some studies suggesting that all subtypes of repetitive behaviours were associated with sensory hyper-sensitivity but not sensory hypo-sensitivity or sensory seeking (Boyd et al., 2010), other studies finding that hyper-responsiveness was associated only with the insistence on sameness behaviours but not motor stereotypies (Baranek et al., 1997) and that hypo- rather than hyper-sensitivity was associated with motor behaviours (Gal et al., 2010). It is important to stress the fact that these studies used different measures for assessing both repetitive behaviours and sensory problems and differed in terms of sample characteristics which further complicates interpretation of findings Repetitive Behaviours and Anxiety Studies that examined the relationship between repetitive behaviours and anxiety have consistently found that higher levels of anxiety were associated with elevated levels of repetitive behaviours (Sukhodolsky et al., 2008; Rodgers, Riby, Janes, Connolly, & McConachie, 2012; Rodgers et al., 2012b). For example, Rodgers and colleagues (2012) used the RBQ to compare the levels of repetitive behaviours between children with autism who met cut-off criteria for elevated anxiety on SCAS-P and children with autism who did not meet the criteria. Children in high anxiety group had higher total repetitive behaviours, sameness behaviours/circumscribed interests, and sensory-motor behaviours than children in the low anxiety group Sensory Problems and Anxiety The relationship between sensory problems, more specifically sensory hypersensitivity and anxiety have been found in general population (Aron & Aron, 1997) and it is not surprising that this relationship has been consistently replicated in population with autism (Green et al., 2012; Ben-Sasson et al., 2008; Mazurek et al., 2013; Pfeiffer et al., 2005). For example, Pfeiffer et al. (2005) examined the relationships between sensory problems and the affective disorders in a group of 50 children and adolescents aged 6-17 years diagnosed with Asperger s syndrome by using the Sensory Profile and the Revised Children s Manifest Anxiety Scale Adapted-Parent s Version (RCMAS; Reynolds & Richmond, 2000). They

68 68 found a relationship between sensory hypersensitivity, avoidance and anxiety for the entire group as well as for younger (N= 30; age range: 6-10 years) and older (N= 20; age range: years) children. There was also a significant relationship between the sensory hyposensitivity and symptoms of depression for the whole group and in the older age group but not in the younger group Summary Results of the studies reviewed above seem to provide some support for the arousal theory of repetitive behaviours. More specifically, studies have found that repetitive behaviours were associated with sensory modulation problems, e. g. either hyper- or hyposensitivity. Furthermore, several studies have found an association between repetitive behaviours and anxiety. Finally, anxiety was found to be associated with sensory hypersensitivity in all studies that explored the relationship between anxiety and sensory processing. However, none of the studies have looked at the three-way relationship between repetitive behaviours, sensory problems and anxiety. Chapter 6 of this thesis will examine the inter-relationship between these behaviours. Furthermore, this chapter will examine our proposal (Leekam Prior, & Uljarevic, 2011) that different subtypes of repetitive behaviours are differently associated with anxiety and sensory problems

69 69 Chapter 2: General method Section 2.1 Background This chapter provides details of the participants and measures for this thesis. Both primary and secondary datasets were used. The primary data set was based on the data collected by the author for the Children s and Parents Behaviour Study, a research project carried out in the local region (South Wales). For this project, parents completed a number of questionnaires about their children. The majority of questionnaires related to children s behaviour. These included questionnaires selected to study repetitive behaviours, sensory processing and anxiety. As discussed in Chapters 7-9, parents also completed questionnaires about themselves. Measures relevant to parents included questionnaires on anxiety, coping and other parents variables. In order to carry out more extensive analysis of data specifically on repetitive behaviours, it was possible to access two additional secondary datasets. These are detailed in the sections 2.4 and These secondary datasets gave the opportunity for new analyses to be conducted for the first time to answer specific questions raised by the thesis. The first section of this chapter will provide details on the measurement selection process and the detailed description of the measures used in this thesis. The second section will provide a description of the recruitment process and primary as well as secondary datasets used in this thesis Measures As described in the thesis statement and in Chapter 1, this thesis had two main aims: (1) to explore the relationship between repetitive behaviours, sensory problems and anxiety in children and adolescents with autism; and (2) to explore the prevalence of anxiety in parents of children with autism and the key factors associated with their anxiety problems. Such a broad scope dictated the choice of measures. The measures had to provide detailed assessment of behaviours as heterogeneous as repetitive behaviours and sensory modulation problems in children and concepts as complex as coping strategies and intolerance of uncertainty in parents. Also, it was necessary to collect data on levels of anxiety, autism severity, language levels in children and data on anxiety levels, sensory processing, autismrelated traits and levels of support for parents. It is clear that, as mentioned, measures had to provide detailed and valid information on the above mentioned behaviours and traits but at

70 70 the same time, due to the scope of behaviours and traits examined, every measure had to be economical in terms of time and effort on the part of parents and children because, in addition to being concerned with obtaining quality data that would allow the thesis to answer the questions that were set-up at the beginning, equal concern was applied to not overburdening participants. Having this in mind, the decision was to opt for the use of questionnaires as the source of information for this thesis. Questionnaires are the primary source of information in the literature on anxiety, coping, autistic traits, and intolerance of uncertainty in parents. However, in terms of repetitive behaviours and sensory modulation problems, there are plenty of experimental protocols and observational approaches as well as autobiographical accounts that allow researchers to collect data on the phenomenon in question. Although for the pragmatic reasons, questionnaires were the most feasible method for gathering relevant information on the behaviours of interest, the use of questionnaires was also justifiable at a theoretical level. In terms of repetitive behaviours, the goal of this thesis was to examine the structure of this heterogeneous category of behaviours which is not possible to do by using observational measurements. More precisely, laboratory-based observations allow for a detailed assessment and analysis of repetitive motor behaviours, either with or without objects, however, insistence on sameness types of behaviours are difficult to observe in laboratory settings and it has been suggested that these behaviours are better captured by parents reports (Leekam et al., 2011; Honey et al., 2012). Also, the use of questionnaire measures of repetitive behaviours provides information on behaviours across time and across a range of situations rather than just information on the behaviours in a particular setting. In terms of sensory problems, it is clear that the use of carefully designed experimental paradigms coupled with electrophysiological assessments is important in elucidating mechanisms responsible for the atypicalities in the sensory processing observed in this population. However, the main aim of the first part of this thesis was to examine the relationship between repetitive behaviours, anxiety and sensory modulation problems in the population of children with autism, more precisely, to examine an earlier hypothesis by Leekam, Prior and Uljarevic (2011) that different types of repetitive behaviours are differently related to anxiety and sensory problems. Since this hypothesis has not been examined in the literature before, it was decided that this relationship should be first examined by using well validated questionnaire measures that provide detailed information on each of the problems and phenomena of interest. As will be discussed in the general

71 71 discussion chapter, it is planned to follow up these findings with experimental work. However, for this thesis, it was necessary to establish this relationship using measures that collect information on the presentation of these behaviours and problems across different contexts and situations rather than in one specific and not necessarily ecologically valid context Selection of the questionnaires Parallel with the systematic search of literature on the repetitive behaviours, sensory modulation problems and anxiety in children with autism, and the literature on the anxiety, and coping in parents of children with autism but also on anxiety and related phenomena in general, all the papers were screened for the measures that were used. The following section ( ) will provide an overview of the questionnaires that were used for the studies presented in this thesis. Before describing the questionnaire chosen to assess behaviour/trait/phenomenon of interest, a short description of the selection process will be presented. In general, a similar set of criteria was used when choosing every measure: a.) that it was previously used in the population of children and adolescents with autism (or population of parents/or relevant research are-coping and anxiety in general population); b.) that it was validated in the relevant population; c.) that it was not time consuming yet providing comprehensive assessment of the behaviours/trait/phenomena of interest. It is important to note here that the selection process of the repetitive behaviours measurement is described in a detailed manner in Chapter 3 and the selection process of the anxiety measure is described in Chapter 4. Currently, there is only one questionnaire that measures intolerance of uncertainty-the Intolerance of Uncertainty Scale (IUS; Freeston et al., 1994; Buhr & Dugas, 2002; Carleton, Norton, & Asmundson, 2007) and therefore the selection process for this measure will not be described Description of questionnaires Questionnaires related to children s behaviours (a) Repetitive Behaviours Measure Selection process: as noted above, a detailed description of the selection process is provided in chapter 3, section 3. 2.

72 72 The Repetitive Behaviour Questionnire-2 (RBQ-2; Leekam et al., 2007) is a 20 item parental questionnaire designed to record repetitive behaviours in both children with autism and in typical development. Details on the design of the RBQ-2 will be provided in Chapter 3. Parents are asked to rate the frequency with which their child has engaged in a range of RRBs over the last month. The RBQ-2 has previously been used in a population of typically developed children (Leekam et al., 2007; Arnott et al., 2010). In an exploratory factor analysis conducted on a sample of 590 typically developing 2-year olds, Leekam and colleagues identified that both 4 factor and 2 factor solutions provided good fit for the data. A four-factor model accounted for 51% of the variance with the following 4 factors: repetitive motor movements, rigidity/adherence to routine, preoccupations with restricted patterns of interest and unusual sensory interests. A two-factor model accounted for 39% of the variance with the following 2 factors: the Motor/Sensory and Rigidity/routines/preoccupation with restricted patterns of interests. It has been found that RBQ-2 has good psychometric properties in a population of 15 month old (Arnott et al., 2010) and 24 month old (Leekam et al., 2007) typically developing children. Internal consistency in the 15 month old sample was:.85 for the total RBQ-2 score;.81 for motor/sensory and.71 for rigid/routines/preoccupations subscales (2 Factor solution);.82 for repetitive motor movements,.74 for rigidity,.64 for preoccupations with restricted patterns of interest and.51 for unusual sensory interest subscale (4 factor solution). In the 24 month sample, internal consistency of RBQ-2 was: 85 for the total RBQ-2 score;.80 for repetitive motor movements,.75 for rigidity,.72 for preoccupations with restricted patterns of interest and.66 for unusual sensory interest subscale (4 factor solution). (b) Sensory Processing Measure Selection process: articles identified through the initial systematic literature search on the sensory processing in autism were screened for the measures used. The following sensory processing measures were identified: the Sensory Profile (SP; Dunn, 1999), the Short Sensory Profile (SSP; Dunn, 1999), the Sensory Experiences Questionnaire (SEQ; Baranek, Favid, Poe, Stone, & Watson, 2006), the Sensory Sensitivity Questionnaire-Revised (Talay-Ongan & Wood, 2000), the Diagnostic Interview for Social and Communication Disorders (DISCO; Wing et al., 2002; Leekam et al., 2007), and the Sensory Processing Measure (SPM; Glennon, Miller-Kuhaneck, Henry, Parham, & Ecker, 2007). The following selection criteria were applied: (a.) that a measure was previously used and validated in the population of

73 73 children and adolescents with autism and validated in this population; (b.) that it was appropriate for a wide age range that was used in this thesis; (c.) that it was not time consuming but that it provided a comprehensive assessment of the sensory processing in children and adolescents with autism. More precisely the instrument had to provide an assessment of sensory processing across different sensory modalities and also provide an assessment of sensory modulation problems as defined by the Interdisciplinary Council on Developmental and Learning Disorders (ICDL, 2005) i. e. sensory hyper-, hypo-sensitivity and sensory seeking; (d.) that the instrument was norm referenced. The evaluation of the identified instruments against the set criteria is presented in the table As can be seen, although the Sensory Profile can be considered to be a time-consuming instrument (125 items), it was the only questionnaire that met other criteria and therefore it was chosen for this thesis. It is important to note that since studies that are included in this PhD started in June 2010, it was not possible to consider the inclusion of questionnaires that were published later. One such questionnaire is for example the Glasgow Sensory Questionnaire (Robertson & Simmons, 2013). This questionnaire provides information on sensory processing across different modalities and also provides sensory modulation scores, however, the norms have not been published yet. Table Selection of the sensory processing measures Used & validated in Population with autism Wide age range Not time consuming Comprehensive assessment Norm referenced Sensory Profile (SP; Dunn, 1999) X Short Sensory Profile (SSP; Dunn, 1999) X Sensory Experiences Questionnaire (SEQ; Baranek, Favid et al., 2006) X X Sensory Sensitivity Questionnaire- Revised (Talay- Ongan & Wood, 2000) X X X DISCO (Leekam et al., 2007) X X Sensory Processing Measure (SPM; Glennon et al., 2007) X X X

74 74 The Sensory Profile (SP; Dunn, 1999). SP is a 125 item, norm-referenced, parental questionnaire. SP is divided into three main sections: sensory processing section (further subdivided into 6 subsections that assess sensory processing across different modalities e. g. auditory, touch processing), modulation section (further subdivided into 5 subsections) and behavioural and emotional section (further subdivided into three subsections reflecting emotional and behavioural responses to sensory stimuli). Dunn (1997; 1999) also conducted PCA and identified the following 10 factors: Sensory Seeking, Emotionally Reactive, Low Endurance/Tone, Oral Sensitivity, Inaattention/Distractibility, Poor Registration (or hyposensitivity), Sensory Sensitivity, Sedentary, Fine Motor/Perceptual, and Other. Also, Dunn proposed a model which suggests that the pattern of individual s sensory responsiveness can be classified based on their neurological threshold and behavioural responses to incoming stimuli. The neurological thresholds continuum ranges from high (when high intensity stimuli is necessary in order to respond), to low (when low intense stimuli provokes a response). Based on the interaction between neurological thresholds and behavioural responses, Dunn's model classified the pattern of individual s sensory responsiveness into four quadrants: (1) Sensation Seeking; (2) Sensation Avoiding; (3) Sensation Sensitivity; and (4) Low Registration (or hypo-sensitivity). (c) Anxiety Measure Selection process: as noted above, a detailed description of the selection process is provided in the chapter 4, section The Spence Anxiety Scale-Child Version (SCAS-C) is a self-report questionnaire designed to measure anxiety symptoms in children and adolescents. It is a 44-item questionnaire and the frequency of each item is rated on a 4-point Likert scale (from 1= never to 4= always). The SCAS-C provides a total anxiety score and scores for the following 6 anxiety subscales: panic attacks and agoraphobia (9 items), separation anxiety (6 items), physical injury fears (5 items), social phobia (6 items), obsessive-compulsive (6 items), and generalized anxiety disorder (6 items). Factor analysis by Spence (1997) confirmed a 6 factor structure of SCAS-C. Subsequent studies (Spence, 1998; Spence et al., 2001; Essau et al., 2002) provided further support for proposed subscales, however, 6 factors did not receive support from studies that used the SCAS-C translations (the German translation: Essau, Sakano, Ishikawa, & Sasagawa, 2004; the Chinese: Essau, Leung, Conradt, Cheung, &

75 75 Wong, 2008; the Japanese: Essau et al., 2004). SCAS-C has been shown to have good psychometric properties in the general population. Internal consistency is.92 for the total scale score and ranges from.60 to.82 (Spence, 1997). Six months test-retest reliability was found to be acceptable for both total scale and subscales (Spence, 1998). The Spence Anxiety Scale-Parent Version (SCAS-P) consists of 38 items that correspond to the items contained in the child version. For each item, parents indicate the frequency of particular anxiety symptom on a 4-point scale (from 0= never to 3= always). SCAS-P provides total anxiety scores as well as 6 subscales scores that match subscales from the SCAS-C. SCAS-P has been shown to have good psychometric properties (Nauta et al., 2004). The Spence Preschool Anxiety Scale (SPAS) is a parent-completed questionnaire. It provides total anxiety scores and scores for the following anxiety subscales: separation anxiety, physical injury fears, social phobia, obsessive compulsive disorder, and generalized anxiety disorder. It also contains an additional open-ended item which relates to the child s experience of a traumatic event meant for clinical practice but not scored. The SPAS has been shown to have good psychometric properties in general population (Spence et al., 2001). Multidimensional Anxiety Scale for Children (MASC; March et al., 1997): MASC is a 45-item measure of anxiety for children 4 19 years of age. It has both parental and a child version. MASC provides the total anxiety score and also the scores for the following factors: somatic/panic, general anxiety, separation anxiety, social phobia, and school phobia. Internal consistency (Cronbach s alpha) for these four subscales ranged from.74 to.85 (March et al., 1997). It has been shown to have good psychometric properties in non-asd populations (Wood et al., 2002). However, although it has been used in 6 studies examining anxiety in ASD (Bellini, 2004; 2006; Sze & Wood, 2008; White & Roberson-Nay, 2009; White et al., 2009; Wood et al., 2009) its psychometric properties in this population haven t been evaluated. (d) Autism Severity Measure Selection process: due to the nature of the studies for this thesis, only questionnaire measures were considered. Interview based assessments such as the Diagnostic Interview for Social and Communicative Disorders (Wing et al., 2002), the Autism Diagnostic Interview- Revised (Lord, Rutter, & Le Couteur, 1994), the Developmental, Dimensional and Diagnostic Interview (Skuse et al., 2004) and observational measures such as the Autism Diagnostic

76 76 Observation Schedule (Lord, et al, 2000) could not be conducted. The following autism screening questionnaires were identified through the systematic search of the literature: the Childhood Autism Rating Scale (Schopler, Reichler, DeVellis, & Daly, 1980), the Autism Spectrum Quotient (AQ; Baron-Cohen, Wheelwright, Skinner, Martin, & Clubley, 2001), the Social Communication Questionnaire (Rutter, Bailey, & Lord, 2003) the Social Responsiveness Questionnaire (Constantino et al., 2004), the Childhood Asperger Syndrome Test (Williams et al., 2005). The same set of selection criteria as described in the section was used with an additional requirement that in addition to providing overall autism severity score, the instrument had to also provide subscores for the diagnostic triad described by DSM-IV and ICD-10 criteria i. e. qualitative impairments in social interaction, communication and the presence of restricted, repetitive and stereotyped patterns of behaviour. The only questionnaire measure meeting the last criteria was the Social Communication Questionnaire (SCQ; Rutter, Bailey, & Lord, 2003) The Social Communication Questionnaire (SCQ; Rutter et al., 2003). SCQ (formerly Autism Screening Questionnaire, ASQ) is a 40-items, parental questionnaire developed from the Autism Diagnostic Interview-Revised (ADI-R; LeCouteur et al., 1989). SCQ provides scores for the following 3 subscales: problems in the social interaction, language and communication and repetitive behaviours. In the original study, Berument et al. (1999) reported 15 to be an optimal cut-off score for differentiating between individuals with and without autism (sensitivity=.85 and specificity=.75) and between Individuals with autism from individuals with intellectual disability but without autism (sensitivity=.96 and specificity=.67). However, subsequent studies suggested 11 as a more sensitive cut-off score that improves discriminant validity (Allen et al., 2007; Corselo et al., 2007; Lee et al., 2007; Snow & Lecavalier, 2008) Questionnaires related to parents behaviours and traits In this section, I report the selection of measures against a set of general criteria. (a) Anxiety Measure Selection Process: Only questionnaires designed to specifically assess anxiety (and preferably depression as well) were eligible for consideration due to a more detailed assessment offered by these instruments. Therefore the general instruments such as the

77 77 General Health Questionnaire (GHQ; Goldberg et al., 1978) were not eligible. Furthermore, interview assessments such as the Anxiety Disorders Interview Schedule (Brown, DiNardo, & Barlow, 1994; 2001) were also not eligible. The following anxiety questionnaires were identified through the systematic literature search: the Hamilton Anxiety Scale (Hamilton, 1969), the Clinical Anxiety Scale (Snaith, Baugh, Clayden Husain, & Sipple, 1982), the Spielberger State Trait Anxiety Inventory (Spielberger, Gorrusch, Lushene, Vagg, & Jacobs, 1983), the Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983) and the Beck Anxiety Inventory (Beck, Epstein, Brown, & Steer, 1988). Identified questionnaires were evaluated against the following criteria: (a.) that it was widely used and validated in the general and various clinical populations, across different age groups; (b.) that it was also in a population of parents of children with autism (or at least other neurodevelopmental disorder); (c.) that it was not time consuming and (d.) that it preferably assessed depression in addition to anxiety. As can be seen from Table 2. 2., only the Hospital Anxiety and Depression scale satisfied all of the criteria as none of the other measures offered assessment of depression in addition to anxiety. Although Beck Inventory in addition to anxiety has also a depression inventory version, this is a separate questionnaire whose addition would make the completion of two questionnaires time consuming. Table Selection of the parents anxiety measure Hamilton Anxiety Scale (Hamilton, 1969) Clinical Anxiety Scale (Snaith et al., 1982) Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983) Beck Anxiety Inventory (Beck & Steer, 1993) Used & validated in general and clinical populations Used in population of parents of children with autism Not time consuming Assessing depression in addition to anxiety X X X X

78 78 Hospital Anxiety and Depression Scale (HADS, Zigmond & Snaith, 1983). HADS is a 14-item self-report norm-referenced questionnaire. HADS provides anxiety and depression scores (each of subscales consists of 7 items). The items related to symptoms that might have physical cause (such as weight loss and insomnia) are not included and HADS is considered to be unaffected by potential co-morbid conditions (Snaith & Taylor, 1985). HADS has been shown to have good psychometric properties. (Herman et al., 1997) and to have a good agreement with other self-report measures of psychological distress and with clinical diagnosis (Bjelland, Dahl, Haug, & Neckelmann, 2002). HADS provides cut-off scores for classifying individuals into those who do not exhibit symptoms of anxiety/depression, those who have elevated levels of anxiety/depressions and those with clinically significant levels of anxiety/depression. (b) Coping Measure Selection Process: A systematic search of coping instruments that have been used in research on coping in parents of children with autism as well as parents of children with other neurodevelopmental conditions and various chronic neurological and somatic disorders was conducted. The COPE Inventory (Carver, Scheier, & Weintraub, 1989) and the Ways of Coping Checklist-Revised (Vitaliano, Kusso, Carr, Maiuro, & Becker, 1985; Folkman & Lazarus, 1988) were the only two coping instruments that were widely used in the coping research in the above listed populations. The Cope Inventory had an advantage in terms of number if items when compared to the Ways of Coping Checklist revised (28 vs 66 items). However, the Ways of Coping Checklist was chosen as it has been validated more extensively and used much more widely in a general population and in individuals with anxiety and depression. The Ways of Coping Checklist-Revised (WCC-R; Vitaliano et al., 1985; Folkman & Lazarus, 1988). The WCC-R is a 66 items, self-report questionnaire. The WCC-R provides detailed assessment of escape-avoidance coping referring to individual s efforts to escape/avoid problem by either engaging in wishful thinking and/or active behavioural efforts to escape the problem; and problem-solving coping referring to individual s effort on solving the problem by careful analysis of the problem/situation and generation of possible ways to address/solve the problem/situation). The WCC-R has been used with various non-clinical as well as with clinical populations e. g. with individuals with cancer (Wonghongkul, Moore, Musil, Schneider, & Deimling, 2000; Rosberg, Edgar, Collet, & Fournier, 2002;) multiple

79 79 sclerosis (Jean, Paul, & Beatty, 1999; Lundqvist & Ahlström, 2006) Parkinson's disease (Sanders, Dewey, Mullins, Chaney, 2001), fibromyalgia (Cronan, Serber, Walen, & Jaffe, 2002), stroke (Rochette & Desrosiers, 2002; King, Zeldow, Carlson, Feldman, & Philip, 2002), chronic fatigue syndrome (Ax, 1999), spinal cord injury (Winemann, Durand, & McCulloch, 1994; Change, Lee & Lieh-Mak, 2000), muscular dystrophies and postpolio syndrome (Lundqvist & Ahlström, 2006) and their families. (c) Intolerance of Uncertainty The construct of intolerance of uncertainty, its relation to anxiety and other psychopathology and relevance to this thesis is discussed in chapters 7-9. The Intolerance of Uncertainty Scale (IUS; Freeston et al., 1994; Buhr & Dugas, 2002). IUS is a 27-item, self-report questionnaire originally developed by Freeston et al. (1994). Freeston and colleagues originally found that IUS had a 5 factor structure: (1) uncertainty must be avoided, (2) being uncertain negatively influences the person, (3) uncertainties are frustrating, (4) uncertainties are stressful and anxiety provoking, and (5) uncertainty prevents one from acting. However, subsequent studies failed to find a support for the 5 factor structure of IUS. A recent factor analysis of the IUS (Carleton et al., 2007) suggested that reduced version of the scale (consisting of 12 items) had better psychometric properties that the original 27-ite scales. McEnvoy and Mahoney (2011) recently reviewd factor analytic studies of the IUS conducted thus far, and found that the Carleton et al. (2007) two factor model had the best properties. Due to these reasons this thesis will use Carleton et al s 12 item modification of the Freeston et al. (1994) scale. Several facto analytic studies (Carleton et al., 2007; McEvoy & Mahoney, 2011; Helsen, van den Bussche, Vlaeyen, & Goubert, 2013) found that IU scale consists of two factors-the first termed as Desire for Predictability and the second termed as Uncertainty Paralysis and these factors will be explored in this thesis. (d) Sensory Processing Measures Selection Process: There are currently only two questionnaires suitable for assessing sensory processing in adults-the Adult Sensory Profile (ASP; Brown & Dunn, 2002) and the Highly Sensitive Person Scale (HSP; Aron & Aron, 1997). Since sensory processing in parents of children with autism has never been assessed before and also due to the fact that

80 80 these two instruments differ in terms of their focus, it was decided that both questionnaires would be used for this thesis. The Highly Sensitive Person scale (HSP; Aron & Aron, 1997). HSP is a 20 item selfreport questionnaire designed to measure sensitivity to various stimui, situations and contexts. The measure has been found to have excellent psychometric properties (Aron & Aron, 1997; Smolewska, McCabe, & Woody, 2006; Liss et a., 2007). Aron & Aron (1997) originally proposed that the concept of highly sensitive person should be viewed as unidimensional. However, in contrast to Aron and Aron s finding that the scale is unidimensional, three recent factor analyses (Smolewska et al., 2006; Evans & Rothbart, 2008; Liss et al., 2010) found the following three factors: Aesthetic Sensitivity (AES), Low Sensory Threshold (LST), and Ease of Excitation (EOE). It was hypothesized that LST and EOE represent negative aspects of sensory processing while AES represents positive aspects. Adults Sensory Profile (ASP; Brown & Dunn, 2002). ASP is a 60-item self-report questionnaire. ASP is based on the Sensory Profile and it provides the following 4 quadrant scores: poor sensory registration (or hypo-sensitivity), sensory seeking, sensory sensitivity and sensory avoidance that correspond to quadrant scores provided by Sensory Profile. It has been found to have good psychometric properties. (e) Empathy Traits Measure (proxy for Broader Autism Phenotype Measure) Selection Process: rather than using the Autism Spectrum Quotient (AQ; Baron- Cohen et al. 2001), the Social Responsiveness Scale-Adult (SRS-A; Constantino & Todd 2005) or the Broad Autism Phenotype Questionnaire (BAPQ; Hurley et al. 2007) as well as the established measures of the BAP traits, the decision was made to concentrate on the cognitive and affective empathy as proxy for the BAP traits. The reasons for this were twofold. Firstly, impairments in cognitive and affective empathy are considered to be a part of the BAP (Sucksmith, Allison, Baron-Cohen, Chakrabati, & Hoekstra, 2013) and more importantly, they have been previously found to be associated with anxiety in general population. Thus, there were strong theoretical reasons to concentrate on cognitive and affective empathy as features of BAP that might be associated with anxiety in parents. The second reason for concentrating on cognitive and affective empathy rather than using above mentioned measures was a pragmatic one as initial feedback received from parent support groups suggested that parents were not likely to complete classic BAP screening measure and that in some cases it might cause additional stress in parents. The Interpersonal Reactivity

81 81 Index (IRI; Davis, 1994) was chosen as a well established measure of various aspects of empathy. The Interpersonal Reactivity Index (IRI; Davis, 1993). IRI is a 28 item, self-reported questionnaire. IRI provides total scores as well as scores for the following 4 subscales: Perspective Taking or Cognitive Theory of Mind, Empathic Concern or Affective Theory of Mind, Personal Distress (defined as tendency to experience distress or discomfort in response to extreme distress in others; ) and Fantasy (defined as the tendency to imaginatively transpose oneself into fictional situations ). Higher score in a subscale represents a higher functioning in each aspect of empathy. IRI subscales have been shown to have good psychometric properties (Davis, 1980; Siu & Shek, 2005) Background questionnaire In addition to the above described questionnaires, a background questionnaire was also used. It collected data on the type of child s diagnosis, date of the diagnosis and the diagnostic centre that established the diagnosis, the presence of co-morbid conditions in children, type of treatment that children receive for their autism related problems and also about the type of treatment that children receive for co-morbid condition, whether the child has a statement of the special education needs and whether the child receives extra support at school, the level of expressive language of child and whether child uses reciprocal social communication or not. The background questionnaire also collected information on parents: their educational level, nationality and language used at home. The ackground questionnaire is provided in the appendix 4. As Expressive Language measure from the Background Questionnaire will be used frequently throughout this thesis, a short discussion on this measure will be provided below. Parents completed this item as a part of the Background Questionnaire. Please see Table 2.3 for the overview of the Expressive Language measure.

82 82 Table Expressive language measure adapted from DISCO What kind of language does your child use? No speech or sounds 0 Gives the names of some people or things when asked 1 Spontaneously say names of several familiar objects for some purpose 2 Says phrases not longer than three words 3 Say some longer phrases (longer than 3 words), missing out the small linking words (e.g. when time go on holiday? ) Talk in spontaneous sentences using small linking words 5 Use past, present and future tenses in complex sentences (e.g. perhaps I will go out tomorrow if it has stopped raining ) Score 4 6 The Expressive Language measure was taken from the Diagnostic Interview for Social and Communication Disorders (DISCO; Wing et al., 2002; Leekam et al., 2002). The DISCO itself has been validated for the use in research (Wing et al., 2002; Leekam et al., 2002; Kent, Carrington et al., 2013). Although this expressive language subscale has been independently used in published research (Honey et al., 2007) and also in unpublished PhD thesis (Honey, 2007), it has not been validated prior to this thesis. The current work provides an initial validation of the Expressive Language DISCO subscale as this scale was significantly correlated with WASI Verbal IQ (r=.558, p=. 047), WASI full IQ (r=.589, p=.044), Mullen s Scale of Early Learning Expressive Language Score (r=.868, p=.005) and Mullen s Receptive Scale Score (r=.787, p=.021) in the subsample of children and adolescents with autism from the primary sample (please see page 83 for detailed description of this sample). However, the sample size of children that were tested using cognitive assessments was small with 12 children tested on WASI and 8 children tested on Mullen s assessments. An independent published dataset of DISCO data (N = 88) was consulted in order to try to validate the language expressive scale used in this thesis. This analysis used the dataset from Kent, Carrington et al. (2013) and this particular analysis is reported in Kent, 2013 (PhD thesis). Using this independent dataset (children of the same age range selected), the expressive language scale was compared with the WASI verbal IQ score. A correlation of r =.467, p =.000 was found. Although in this thesis I used the expressive language measure alone, this result indicates that it is comparable to other measures (more on the limitations related to cognitive testing will be presented at the end of this chapter). Therefore, despite the

83 83 initial validation, the use of this scale will present an important limitation and all analysis where this scale is used are exploratory in nature and results should be considered as preliminary Recruitment process Ethics Ethical approval for all the studies using the primary sample (South Wales) was given by the Research Ethics Committee of the School of Psychology, Cardiff University and by two local NHS health boards Recruitment Procedure Families were recruited through local schools with specialist autism provision and parent support groups. Recruitment centred on the Cardiff and Vale area of South Wales. The recruitment period continued across a period of two years. Head teachers of schools and co-ordinators of autism parent support groups were initially contacted about the project via or post. An or letter was sent that provided the background of the study, outlined the procedures for data collection and also included a copy of questionnaire booklet that families would receive. Head teachers and parent support group co-ordinators who expressed interest to help with the recruitment for the research project were contacted by telephone to discuss the research further. In cases where schools organized parents meetings, a talk was given to a mixed group of parents and teachers, outlining the background of the research project. Talks were also given to parent support groups outside of school settings. After learning more about the study, families were provided with leaflets that, in addition to brief description of the study, contained contact details of the PhD student, and told to contact the PhD student in case they were interested in taking part. Upon contacting, potential participants were again provided with a brief description of what the study involves and in case the participants were still interested in taking part in the study, an initial phone interview was scheduled. During a brief phone interview (approximately 5 minutes per participant), families were screened for eligibility of their child to take part in the study. Inclusion criteria were the following: the child, having received a formal diagnosis of autism by a professional (e.g., psychologists, psychiatrists, neurologists with experience working with individuals with autism), aged from 2 to 17 years, and absence of epilepsy, brain injury, cerebral palsy, any neuro/musculo/skeletal disorder/malformation that would seriously limit

84 84 ability to walk without help or a known genetic condition (e.g., Fragile X, Down syndrome). The child s language comprehension was also screened in order to choose the apropriate anxiety measure. After the phone screening, parents were then sent Booklet 1 containing 5 questionnaires. Please see tables for a list of questionnaires included. Schematic representation of the whole recruitment process and the steps involved in the distribution of questionnaires and phone interviews is summarized in the figure below. Schools/parent support groups contacted Talk to parents given, parents expressed interest in taking part Booklet 1 sent to participants Initial contact with families: screening for eligibility, age and language level of the child Participants completing the booklet 1 (2 weeks) Participants contact the researcher if booklet is completed/particip ants contacted after 2 weeks 1. Phone interview (Background Questionnaire) 2. Participants asked if they would like to receive booklet 2 Participants contacted after receiving the questionnaire Participants completing the booklet 2 (2 weeks) Booklet Received Figure Schematic representation of the recruitment process

85 85 Table Questionnaires (about the child) included in the Booklet 1 Behaviour/Trait Measured Repetitive Behaviours Sensory Processing Anxiety Questionnaire (completed by) Repetitive Behaviour Questionnaire-2 (Parent) Sensory Profile (Parent) Spence Anxiety Scales (Parent and child) Multidimensional Anxiety Scale for Children (Parent) Autism Severity Social Communication Questionnaire (Parent) During their initial phone call, parents were asked to contact the researcher within two weeks after receiving a questionnaire and also they were asked for a permission to be contacted by the researcher if he didn t hear from them. During the second phone call with parents, the researcher went through the background questionnaire with parents (described in the measures subsection) and asked if they would like to receive the booklet 2 containing additional 6 questionnaires. Please see Table for a list of questionnaires included. The second phone interview lasted between 15 and 20 minutes. Booklet 2 was sent to families who agreed to continue with the research. Table Questionnaires (about parent) included in the Booklet 2 Behaviour/Trait Measured Anxiety Questionnaire (completed by) Hospital Anxiety and Depression Scale (Parent) Empathy traits Coping Intolerance of Uncertainty Sensory Processing Interpersonal Reactivity Index (Parent) Ways of Coping Scale (Parent) Intolerance of Uncertainty Scale Adult Sensory Profile Highly Sensitive Person Scale It is important to note here that this PhD research was a part of a larger research project conducted by the Wales Autism Research Centre. In addition to research questions

86 86 that were specific to this PhD, this project also explored various aspects of social and communication difficulties in children and adolescents with autism and also explored repetitive behaviours from a functional analytic perspective and due to this, additional questionnaires that were not used in this PhD project were also collected (9 additional questionnaires). As a result of this, parents had to complete 20 questionnaires in total and a phone interview. This required at least 3 hours of their time and because of the time consuming nature of this project and necessity to conduct an interview over the phone, it was not possible to collect data on-line. It was also felt that because parents had to complete a series of questionnaires about their own anxiety, coping strategies and other, potentially sensitive and stress inducing topics, it was necessary to present details about the study to parents in the form of a talk given by PhD student and allow parents to ask any question/express concerns related to the project. This also precluded online recruitment and data collection Description of the primary sample (South Wales Sample) Although parents also provided some data about themselves for this section, the results are predominantly based on their children s behaviours and characteristics. Therefore, the group characteristics presented are about children, as reported by their parents. Sixty three parents of children with DSM-IV-TR diagnoses of ASD returned completed Questionnaire Booklet 1 and 59 returned booklet Questionnaire 2. Data for the Group with autism was primarily provided by biological mothers, although two biological fathers provided data. In this section I will provide descriptive data for the whole sample of children in terms of their chronological age, expressive language levels, co-morbid diagnosis, nationality, their parents educational level and language spoken at home. Data for all the studies presented in this thesis will be drawn from this sample. Descriptives for the specific relevant samples will be presented in the relevant chapters. Before providing descriptives for the whole sample it is important to note that from the whole sample (N= 63), 59 provided data on RBQ-2 questionnaire and this subsample was used in Chapter 3. As the goal of Chapters 4, 5 and 6 was to examine interrelationship between repetitive behaviours, sensory problems and anxiety, only individuals for whom data were available for all three measures were included in analyses conducted for these 3 chapters. As both the Sensory Profile and the Spence Anxiety Scales are norm-referenced measures, they do not allow for any missing data which reduced the size of sample used in Chapters 4, 5 and 6 to N= 49. Finally, aim of chapters 8

87 87 and 9 was to examine frequency and risk factors for anxiety in mothers of children with autism. As the Hospital Anxiety and Depression scale and the Adult Sensory Profile are both norm-referenced instrument, they also do not allow for any missing data and size of the sample for Chapters 8 and 9 was N= 50 (out of those 50 cases 43 overlapped with sample used in Chapter 4, 5 and 6; this was a result of a combination of missing data from sample used in Chapters 4, 5 and 6 in terms of mother s anxiety and sensory questionnaires and data missing from sample used in Chapters 8 and 9 in terms of children s anxiety scales which resulted in 7 non-overlapping cases). Diagnosis: Each child had a clinical diagnosis. As this was a questionnaire study and the time commitment for each participant was already high, it was decided not to reconfirm diagnosis for each child by the use of diagnosis tools such as the ADI-R (Lord, Rutter, & LeCouteur, 1994; Lord et al., 2000), ADOS (reference) or DISCO (Leekam et al., 2002; Wing et al., 2002). However, clinicians from two health boards (Cardiff and Vale and Aneurin Bevan), involved in diagnosing children, collaborated on this project. They confirmed the use of DSM-IV-TR diagnosis and the routine use of ADOS (following the ICD-10 criteria) by these services. Parents reported details of their child s diagnosis % of children had a clinical diagnosis of Asperger s Syndrome, 42.4% of Autism, 25.4% of Autism Disorder and 5.1% of High Functioning Autism. Social Communication Scores (SCQ; Rutter et al., 2003) were available for 43 children who had a developmental level in the appropriate range for this scale. The mean score was 27,63 (SD=5.82), range was Three children scored 14, one scored 16 and 39 scored 21 or above. Several studies (e.g. Eaves, Winger, Ho, & Mickelson, 2006; Allen, Williams, & Hutchins, 2007; Lee, David, Rusyniak, Landa, & Newschaffer, 2007; Wiggins, Bakeman, Adamson, & Robins, 2007) report SCQ scores of 11 or 12 as a valid cut off point to indicate ASD when using this parent completed checklist. Age and language ability: The mean age of children was 9 years 11 months (SD= 4 years 4 months; range: 2 years 5 months to 17 years 9 months; 54 males). Expressive language level estimates were based on a language questionnaire item taken from the Diagnostic Interview for Social and Communication Disorders (Wing et al., 2002) used by Honey, Leekam, Turner, & McConachie (2007). Parents were asked to report if their child had no words, single words, 2-3 words, longer phrases, spontaneous sentences or complex sentences with past, present and future tense. 62.7% of parents reported that their child used complex grammatical speech or spontaneous sentences, 16.6% that their child used phrase

88 88 speech and 21.7% that their child used single words or no speech. 42.2% of children did not have any co-morbid conditions, 22% had one, 22% had two, 10.2% three and 1.6% had 4 comorbid conditions. Parents were asked to indicate if their child had any other health or learning difficulties in addition to autism. Table shows the frequency of specific conditions in this sample of children. Table Health Difficulties in children Health Difficulty Number of children Sleep Problems 12 Gastro-intestinal problems 11 Asthma 8 Alergies 5 Dyspraxia 5 ADHD 5 OCD 1 Dyslexia 1 Parent education: SES data were not available but data on educational level showed that 32% of parents had postgraduate qualifications, a further 30% had undergraduate or vocational qualifications and 25% did not have post-school qualifications (12.3% missing data) Description of the secondary data sets Secondary Data Set 1 Data set 2 consisted of 61 children with autism. This sample consists of archival data of participants recruited from Cambridge University. The original purpose of this project was to assess the relationship between head circumference and cognitive/behavioural correlates related to autism. Only RBQ-2 data were used from this dataset with the full permission from the first author working on this project Dr Jillian Sullivan and her PhD supervisor Professor Simon Baron-Cohen. The mean age of the sample was 5 years 4 months, (SD= 1 year 2

89 89 months; range 2 years 9 months to 8 years 5 months; 56 males). All had a clinical diagnosis of an ASD, established according to ICD-10 criteria. Data set 2 was used only in the first study presented in this thesis. RBQ-2 data from this sample were merged with RBQ-2 data from South Wales sample in order to increase power for the factor analysis of RBQ Secondary Data Set 2 This data set included longitudinal data from 88 typically developing children whose parents completed RBQ-2 questionnaire at when children had 15, 24 and 72 months. The children were part of the Tees Valley Baby Study (Durham University), an opportunity sampled community sample from the North-East of England, studied prospectively via parent questionnaires, observation and interview techniques. The majority of the children were White and British. SES using Hollingshead s (1975) scale represented the full range of deprived to affluent. Ethical permission was granted from Local Research Ethics committees and University Ethics Committees and parents provided written consent for their child to participate in the study. Secondary data set 2 was only used in the first study presented in this thesis with the goal of examining developmental changes of repetitive behaviours in a typically developing general population child sample. The data set is used with the full permission of principal investigator Professor Sue Leekam and the researcher who worked on this study Methodological Limitations of this thesis This PhD was the first study conducted in the Wales Autism Research Centre and because of this no recruitment register existed and no connections with local schools and parent support groups were established before the commencement of this PhD project. Prior to any data collection it was first necessary to establish connections with both local schools and support groups and this involved series of meetings and talks to both teachers and various support and interest groups, more than 25 in total. As a result, recruitment and data collection process was very slow, and spanned over a period of two years. Although it was initially planned to recruit only from Cardiff area, due to slow response it was necessary to widen the recruitment to the whole of Wales. As described above, it was not possible to recruit and collect data online. This resulted in two significant limitations of this PhD. The first limitation, the consequence of a recruitment process, is a relatively small sample size. Because of the nature of recruitment (participants spread all over Wales), only questionnaire

90 90 measures were suitable together with phone interviews. As a consequence of this, it was not possible to conduct cognitive testing on the major portion of the sample (only 12 children were tested on WASI and 8 on Mullen s scale), therefore, there is a lack of a proper index of developmental level in this thesis. Also, as questionnaires were returned by post, there were missing data that rendered norm-referenced questionnaire unusable and further reduced sample size, and although in some cases it was possible to contact parents again regarding the missing data, in majority of cases, due to geographical constraints this was not possible.

91 91 Chapter 3: Repetitive Behaviours in Autism and Typical Development Chapter Plan: In the first part of this chapter, a brief overview of the literature on repetitive behaviours in both autism and typical development will be provided followed by a summary of the existing instruments for measurement of repetitive behaviours. After this, empirical work is provided for both autism and TD. The first aim is to examine the structure and psychometric properties of the RBQ-2 questionnaire in a population of children and adolescents with autism. The second aim is to examine the relationship between repetitive behaviours, chronological age, other core autism symptoms, expressive language and adaptive behaviours. Finally, the third aim is to examine the developmental trajectory of repetitive behaviours in a sample of typically developing children. The results will be discussed in the light of the existing literature Introduction As discussed in more detail in Chapter 1, findings from the literature on the development of repetitive behaviours in autism seem to support the suggestion put forward by Prior & Macmillan (1973) and Turner (1999) that repetitive sensory and motor behaviours can be conceptualized as lower level behaviours as they are generally found in younger children with autism and Individuals with autism with lower developmental level (Moore & Goodson, 2003; Fecteau et al. 2003; Esbensen et al., 2009). However, it is important to note that both sensory and motor behaviours continue to be seen in high functioning groups (South, Ozonoff, & McMahon, 2005). The conceptualization of IS behaviours as higher level behaviours is less straightforward as several studies failed to find association between this factor and developmental level. For example, Lam, Bodfish, & Piven (2008) found that insistence on sameness behaviours were not associated with either age nor verbal IQ in a group of 316 Individuals with autism (mean age= 9.02 years). Also, a study by Murphy, Beadle-Brown, Wing, Gould, Shah, and Holmes (2005) found that RRBs characterized by resistance to change did not change over time. Part of the conflicting results might be the fact that studies mentioned previously were cross-sectional. A longitudinal sstudy by Richler, Huerta, Bishop & Lord (2010) used the ADI-R to track longitudinal change in RRBs in children with ASD over a 7 year period (RRBs were assessed when children were 2, 3, 5, and 9 years of age). Results showed that RSM behaviours remained relatively stable over a 7 year

92 92 period but RSM behaviours did decrease in children with higher NVIQs. Unlike RSM behaviours, IS behaviours, gradually (moderately) increased over time. In addition to older chronological age, higher IS scores were also associated with milder social and communication impairments. However, Richler et al. (2010) did not include typically developing group in their study and also, they did not conduct an analysis of the predictive effect of early RSM behaviours on higher level IS behaviours at later ages. It is of crucial importance to consider findings on the change in repetitive behaviours in autism against what we know about the structure and change of repetitive behaviours during typical development. Several studies have used cross-sectional data to examine the developmental change of RRBs in typical development. For example, Evans et al. (1997) examined the developmental trajectory of 'just right' behaviours, and 'repetitive behaviours & insistence on sameness' using Childhood Routines Inventory (CRI; Evans et al., 1997). Evans and colleagues reported that 'just right' behaviours occurred significantly more in children aged 24 to 48 months than in children aged 12 months and 72 months, while 'repetitive behaviours and insistence on sameness' (e. g. preference for the same daily routine) were present to a similar degree as 'just right' behaviour, although these behaviours emerged earlier and had a steeper increase over time. Arnott et al. (2010) using the RBQ-2 found that the most common behaviour type at the age of 15 months were motor and sensory behaviours. Arnott and colleagues also reported that the total RRB score, especially for the motor total scores and sensory subtype was higher at 15 months than that reported for 2-year-olds. However, to date, there has been no published longitudinal study of RRBs in a typical community sample that describes the development of the range of RRBs and their subtypes across multiple age points. Another issue that has not been addressed in detail is the question of the nature of the relationship between repetitive behaviours and core autism features. Although it is usually assumed that repetitive behaviours are related to the impairments in reciprocal social interaction and deficits in communication, several authors have questioned this assumption (Constantino et al., 2004; Happe, Ronald, & Plomin, 2006; Happe & Ronald, 2008). For example, Constantino et al. (2004), argued that unitary, rather than the three-factor solution provides the best explanation for the clustering of autism symptoms. On the other hand, Happe et al. (2006) and Happe and Ronald (2008) suggested that the association found between the social and non-social impairments is purely coincidental and that these impairments have distinctive genetic etiologies. Studies that examined the relationship

93 93 between repetitive behaviours and social/communication impairments have, in general found that higher RRBs scores were associated with more social/communication problems (Charman et al., 2005; Gabriels et al., 2005: Lam, Bodfish, & Piven, 2008; Ray-Subramanian & Weismer, 2012) For example, Lam et al. (2008) used ADI-R in a sample of 316 children with autism (mean age: 9 years) to look at the relationship between different RRBs subtypes and social/communication deficits. They found that both higher repetitive motor behaviours and insistence on sameness were significantly associated with more social/communication problems. However, no association between circumscribed interests and social/communication deficits was found. To summarize, studies conducted thus far have provided some support for the two factor structure of repetitive behaviours originally proposed by Prior & Macmillan (1975), although several studies have found that 3 or even 5 factors provide a better description of the heterogeneous class of repetitive behaviours. Studies that examined the influence of chronological and developmental age on the expression of repetitive behaviours in autism have provided some support for conceptualizing repetitive sensory-motor behaviours as lowlevel behaviours as these behaviours seem to be more frequent in younger children and individuals of lower developmental level. However, it has also been found that RSM behaviours are frequent in older and high functioning individuals with autism. The results regarding the conceptualization of insistence on sameness behaviours as higher-level behaviours have been even more conflicting. At present, it is not clear whether RSM and IS behaviours are different classes of behaviours or whether, RSM behaviours might predate IS behaviours in developmental terms. Longitudinal studies on RRBs in typical development might provide an answer to this question, however, all the studies that looked at these two classes of behaviours have been cross-sectional. Based on this short review of the literature, it is clear that it is necessary to examine how different classes of repetitive behaviours are related to the chronological age and other core autism features in a systematic way and more importantly, to look at how RRBs change over time during the typical development. This chapter will use secondary RRBs data in order to address such questions. However, as argued in Chapter 1, the type of measure that is used to assess RRBs sets the boundaries for the phenomenon under focus and necessarily influences the conceptualization of RRBs and, before examining any of the above mentioned issues, it is necessary to choose a measure that will allow a detailed and systematic assessment of repetitive behaviours, both in autism and TD population.

94 Measurement of Repetitive Behaviours As reviewed above and in the Chapter 1, repetitive behaviours are a very heterogeneous group of behaviours that are present in the population with autism but also in other neurodevelopmental and psychiatric conditions, and during early development. Hence, it is necessary that every assessment tool must be comprehensive enough in order to gather information on a wide range of behaviours. Because of the above mentioned reasons, the instrument must also go beyond behaviours which may be exclusive to the Population with autism. This would also help to avoid the circularity of simply re-describing diagnostic criteria. However, it is also important to highlight the fact that including a large number of questions may not be suitable for the inclusion in research projects where participants are asked to complete other measures because of the time constraints. Therefore it is important to strike a balance between the need to provide a detailed and comprehensive assessment of the behaviours in question and the reduction in time associated with completion of the measure. To briefly summarize, a strong measure should be: with sound psychometric properties, allow examination of a wide range of repetitive behaviours, both the ones present in autism and the ones present during typical development and in various other conditions, allow the examination of repetitive behaviours across all ages, and finally, be easily administered and not too time consuming. Articles that were identified in a systematic search of the literature for Chapter 1 were also screened for the type of questionnaire/interview measure that was used to assess the repetitive behaviours. Twenty eight questionnaire/interviews were initially identified. However, only the instruments that assess at least two behaviours from each of the four categories of repetitive behaviours described in the current diagnostic criteria for ASD are reviewed here. Despite the fact that Maudsley Item Sheet used by Carcani-Rathwell, Rabe- Hasketh, & Santos (2006) met the initial inclusion criteria, this instrument was not evaluated as it was not possible to obtain the full description of the repetitive behaviours part of the instrument from the published literature. In the end, a total of 12 instruments met the initial inclusion criteria. A detailed description of instruments is provided in Appendix 2 and evaluation of instruments is provided below.

95 Evaluation of the reviewed instruments against the previously set criteria As can be seen from Table 3. 1., RBS/RBS-R questionnaires provide a detailed assessment of various types of repetitive behaviours, however, the time needed for the instrument to be completed might not make it an optimal instrument for the research projects with multiple measures. Similar can be said for DISCO that, as mentioned, is a very detailed assessment interview, however, it is very time consuming. The ADI-R was developed around the DSM-IV criteria for autism which makes it less than a perfect instrument for researching repetitive behaviours across different conditions. Similar can be said about the SCQ that was modelled around ADI-R algorithm items. The CRI and especially Y-BOCS were developed to assess what can be considered as higher order repetitive behaviours and as such do not represent the full spectrum of all types of repetitive behaviours that individuals might show. Table Questionnaire/Interview Evaluation Suitability across Easily Measure administered/not Psychometric Ages Diagnostic groups time consuming Properties SCQ X X DISCO X ADI-R X X Y-BOCS/CY- BOCS X X RBS/RBS-R X RBI/RBQ RBQ-2 (TD) CRI X Only 3 questionnaires (RBI/RBQ/RBQ-2) met all the criteria. I decided to use the RBQ-2 that was developed by our group (Leekam et al., 2007) as this instrument, as mentioned above, provides a detailed assessment of repetitive behaviours, items are applicable across conditions and chronological ages, and since it is a 20 item questionnaire it is not time consuming. In addition, unlike CRI, it covers a wider range of behaviours, while the CRI is more appropriate for the assessment of IS types of repetitive behaviours. Also, the

96 96 RBQ-2 is more applicable across different age range and developmental level than RBS-R which is more applicable to younger children and/or individuals of lower developmental levels. The first objective of this chapter (Study 1) is to explore the developmental trajectory of repetitive behaviours in typical development. The second objective (Study 2) is to assess the factor structure and the psychometric properties of the RBQ-2 in a sample of Individuals with autism with autism spectrum disorders. It is hypothesized that the two or four structure of the RBQ-2 will be confirmed via exploratory factor analysis. The third objective is to examine if repetitive behaviours are moderated by any of the following variables: chronological age, the presence of functional language, communication and social impairments Study 1 Developmental change in RRBs in typical development Methods Subjects Characteristics of the typically developing children were described in more details in Chapter 2 1. Briefly, parents of 88 children completed RBQ-2 at 3 age points: when their children were approximately 15 months (age range: months, mean age: 14.86), 24 months and approximately 72 months (age range: months, mean age: 77.4) old. Measures The Repetitive Behaviour Questionnaire-2 secondary data were used. Results Mean RBQ-2 scores broken down by age are presented in Table Scores for RSM and IS behaviours at both 15 and 24 months have previously been published (Arnott et al., 2010) and are provided here to enable comparison with 6 year old (unpublished) data collected from parents of the same children. RSM and IS subscales were calculated following the procedure used by Leekam et al. (2007) and Arnott et al., 2010). 1 As described in the Chapter 2, this was a secondary data set 2.

97 97 Table Mean RBQ-2 scores at 15, 24 and 72 months Age RSM IS RBQ-2 Total Score 15 months 1.83 (SD=.43) 1.37 (SD=.32) 1.62 (SD=.30) 24 months 1.52 (SD=.40) 1.54 (SD=.42) 1.55(SD=.33) 72 months 1.31 (SD=.36) 1.39 (SD=.37) 1.40 (SD=.31) From Table it is clear that RSM and total RRBs steadily decreased with age while IS behaviour increased between 15 and 24 months and then decreased by 72 months (although mean score was still higher than at 15 months). It is important to note that as most of RBQ-2 subscale scores were skewed, analyses were run with logarithmically transformed data. A series of paired t-tests were performed in order to examine whether there were significant differences between mean RBQ-2 scores at 15, 24 and 72 months. Mean RSM score at 15 months was significantly higher than at both 24 (p=. 000; Cohen s d=.72) and 72 months (p=.000; Cohen s d= 1.31) and that mean RSM score at 24 months was significantly higher than at 72 months (p=.000; Cohen s d=.55). Mean IS score at 24 months was higher than at both 15 months (p=.000; Cohen s d=.49) and at 72 months (p=.01; Cohen s d=.37). Mean RBQ-2 total score at 72 months was significantly lower than at both 15 (p=.000; Cohen s d=.75) and at 24 months (p=.004; Cohen s d=.45). There was no difference between RBQ-2 total scores at 15 and 24 months (Cohen s d=.22). A series of paired T tests was conducted to compare RSM and IS scores at 15, 24 and 72 months 2. It was found that at 15 months, RSM score was significantly higher than IS score (t= 9.977, p=.000; Cohen s d= 1.21), at 24 months there were no significant differences between RSM and IS scores (t=.28, p=.779; Cohen s d=.05) nor were there differences at 72 months after adjusting significance level to avoid Type I error (t= 2.085, p=.04; Cohen s d=.22). Mean IS, RSM and total RBQ-2 scores at three times points, broken down by gender are shown in table A series of mixed RRBs type x gender ANOVAs were performed for 2 Comparison between IS and RSM scores at 15 months was already performed by Arnott et al. (2010) in a larger sample (N= 139) from this data set. These comparison are performed here for the sake of comprehensiveness. However, comparison at other ages nor across agers were not performed previously in any of the published papers and that is an original contribution of this thesis using the secondary dataset.

98 98 15, 24 and 72 months of age. Although, as can be seen from the table 3. 6, males had higher mean RSM, IS and total RBQ-2 scores at all ages, the only differences that reach statistical significance were for RSM (p=.014; Cohen s d=.52) at 72 months and total RBQ-2 scores (p=.013; Cohen s d=.55) at 72 months. As noted above, analyses were run with logarithmically transformed scores. Table Mean RBQ-2 scores at 15, 24 and 72 months for each gender RSM IS RBQ-2 Total Score Age Male Female Male Female Male Female (N= 49) (N= 39) P (N= 49) (N= 39) P (N= 49) (N= 39) P 15 months 1.89 (.47) 1.76 (.38) (.37) 1.34 (.26) (.32) 1.56(.27) months 1.54 (.44) 1.50 (.36) (.44) 1.47 (.38) (.37) 1.49 (.28) months 1.39 (.41) 1.21 (.26) (.43) 1.33 (.27) (.33) 1.31 (.25).013 Linear regression analyses were used to examine the predictive effect of RSM and IS behaviours on the same behaviours at later ages. RSM behaviours at 15 months predicted RSM at both 24 months (F(1, 86)= ; Adjusted R 2 =.350; t= 6.919, p=.000, β=.622) and at 72 months (F(1, 86)= ; Adjusted R 2 =.123; t= 3.631, p=.000, β=.362). When both RSM scores at 15 and 24 months were entered as predictors of RSM scores at 72 months, RSM behaviours at 15 months were no longer significant predictors (t=.550, p=.583, β =.061) but at 24 months were (t= 4.453, p=.000, β=.505). The whole model accounted for 28% of variance. The same procedure was repeated for IS behaviours. IS behaviours at 15 months were significant predictors of IS behaviours at both 24 (F(1, 86)= ; Adjusted R 2 =.225; t= 5.123, p=.000, β=.581) and 72 months (F(1, 86)= ; Adjusted R 2 =.236; t= 5.278, p=.000, β=.495). When both IS scores at 15 and 24 months were entered as predictors of IS scores at 72 months, both IS behaviours at 15 months (t= 3.112, p=.003, β=.309) and at 24 months (t= 3.863, p=.000, β=.384) were significant predictors. The whole model accounted for 34.2% of variance. Next, a set of regressions was performed to examine the relative influence of RSM and IS behaviours as predictors of later outcome. When both RSM and IS behaviours at 15 months were examined as predictors of IS behaviours at 24 months, only IS behaviours were significant predictors (t= 4.205, p=.000, β=.427) while RSM behaviours were not (t= 1.446, p=.152, β=.147). The whole model accounted for 30.6% of variance. When RSM and IS

99 99 behaviours at both 15 and 24 months were entered as predictors of IS behaviours and 72 months, IS behaviours at both 15 (t= 2.379, p=.02, β=.242) and 24 months (t= 3.239, p=.002, β=.332) were significant predictors while RSM behaviours at either 15 (t= 1.611, p=.111, β=.176) or 24 months (t=.623, p=.535, β=.070) were not. The whole model accounted for 36.8% of variance. Similar set of regressions was performed with RSM behaviours as dependent variable. When both RSM and IS behaviours at 15 months were examined as predictors of RSM behaviours at 24 months, only RSM behaviours were significant predictors (t= 5.823, p=.000, β=.540) while IS behaviours were not (t= 1.634, p=. 106, β=.152). The whole model accounted for 36.3% of variance. The whole model accounted for 36.3% of variance. When RSM and IS behaviours at both 15 and 24 months were entered as predictors of IS behaviours and 72 months, only RSM behaviours at 24 months were significant independent predictor (t= 3.935, p=.000, β=.475) while RSM behaviours at 12 months (t=.477, p=.635, β=.056) not IS behaviours at 15 (t=.015, p=.988, β=.002) and 24 months (t=.731, p=.467, β=.082) were not with the whole model accounting for 26.9% of variance. These regression analyses suggest that IS and RSM behaviours are relatively independent classes of behaviours in developmental terms Study 2 Repetitive Behaviours in Autism Methods Subjects Parents of 120 children with Autism Spectrum Disorders (ASD), with ages ranging from 2 years 5 months to 17 years 9 months (mean 7 years 7 months, SD 3 years 10 months) participated as part of their involvement in a research study being carried out in two different parts of the UK (South Wales, n= 59 3, and the South East of England, n = 61 4 ) investigating the association between repetitive behaviour and other factors. The South Wales sample was aged 2 years 5 months to 17 years 9 months (mean age= 9 years 11 months, SD 4 years 4 months; 54 males) and South East England sample was aged 2 years 9 months to 8 years 5 months; (mean age= 5 years 4 months, SD 1 year 2 months; 56 males). As described in Chapter 2, all had a community multidisciplinary team assessment leading to a best estimate 3 Primary Data Set, please see Chapter 2 for details 4 Secondary Data Set, please see Chapter 2 for details

100 100 clinical diagnosis of an ASD (including autism and Asperger syndrome) according to DSM- IV (American Psychiatric Association, 1994) and ICD-10 (World Health Organization, 1993) criteria. Measures The RBQ-2 questionnaire was used. Results Initial analysis using t-tests revealed that the South Wales and the South East of England samples did not differ in their Total RBQ-2 Score (M= 1.99 (SD=.38) vs M= 1.92 (SD=.44); t (118) =.986, p=.33; Cohen s d=.19). Furthermore, although the two groups differed in age, the Total RBQ-2 score was not associated with age (r= -.018, p =.84). Factor Analysis Before conducting a factor analysis, the rate of item endorsement was examined in order to check if any of the items had less than 10% of frequency (in other words, for every item, the occasional to marked response range had to be endorsed by at least 10% or more of the sample). The endorsement rates can be seen in the table Table Response frequencies for each RBQ-2 item 15 or more Never or Rarely 1 (%) One or more times daily 2 (%) times daily (or at least once an hour) Mean (SD) Does your child: 3 (%) 1. Arrange toys or other items in rows or patterns? (.58) 2. Repetitively fiddle with toys or other items? (.83) 3. Spin him/herself around and around? (.73) 4. Rock backwards and forwards, or side to side, either when sitting or when standing? (.79) 5. Pace or move around repetitively? (.85) 6. Make repetitive hand and/or finger (.84)

101 101 movements? 7. Have a fascination with specific objects (.66) 8. Like to look at objects from particular or unusual angles? 9. Have a special interest in the smell of people or objects? 10. Have a special interest in the feel of different surfaces? 11. Have any special objects he/she likes to carry around? (.79) (.82) (.81) (.85) 12. Collect or hoard items of any sort? (.88) 13. Insist on things at home remaining the same? (.78) 14. Get upset about minor changes to objects (.78) 15. Insist that aspects of daily routine must remain the same? 16. Insist on doing things in a certain way or redoing things until they are just right? 17. Play the same music, game or video, or read the same book repeatedly? 18. Insist on wearing the same clothes or refuse to wear new clothes? 19. Insist on eating the same foods, or a very small range of foods, at every meal? (.80) (.82) (.67) (.77) (.85) Since none of the items had frequency of less than 10%, the next step was to conduct a factor analysis. The principal component analysis (PCA) with varimax rotation was chosen following the same methods used by other researchers working in this field (Evans et al., 1997; Szatmari et al., 2006; Leekam et al., 2007; Honey et al., 2012). Item 20, the nonquantitative response item, was excluded. Initial screening showed that assumptions of nonmulticollinearity, sampling adequacy, and factorability were all met (The Kaiser-Meyer- Olkin (KMO) measure verified the sampling adequacy for the analysis, KMO =.779. Bartlett s test of sphericity, χ 2 (171) = , p<.001, indicated that correlations were sufficiently large for PCA).

102 102 An initial analysis was run to obtain eigenvalues for each component in the data. Five components had eigenvalues over Kaiser s criterion of 1 and in combination explained 60% of the variance. However, it has been pointed out that this method has a tendency to retain more factors than is appropriate (Hayton et al., 2004). The other commonly used rule is to use the Scree Test (Figure 3. 1.). Figure Scree Plot However, the criticism of basing the decision on a number of factors that should be retained on just the Scree Plot is that the decision is not robust and reproducible, therefore we used Horn s parallel analysis (PA; Horn, 1965) which is based on the Monte Carlo simulation of random production of Eigen values to determine the number of components that should be extracted. The Parallel Analysis Program was used to run parallel analysis and it was shown that factors 3 onwards have Eigen values less than that from simulations and 2 factor solution should be retained in the final analysis. The PCA with varimax rotation was rerun specifying a two-factor solution. In combination, these two factors accounted for 40.3% of the variance. Factor loadings for items were set at.38 following Comrey & Lee (1992, cited in Tabachik & Fiddell, 2001). Table shows the final 2-factor solution (RSM factor accounted for 11.1% and IS 29.2% of the variance).